Point a telescope at a stellar nursery. Count the stars. Count the brown dwarfs. Keep counting as objects get fainter, cooler, smaller.

At some point, theory says, you should stop finding objects. There should be a mass below which molecular clouds simply can't collapse to form self-gravitating bodies. The prediction comes from opacity-limited fragmentation—dust grains block radiation so efficiently that cloud fragments below a certain mass can't cool and contract.

The cutoff should occur somewhere around ten Jupiter masses, maybe lower. Below that threshold, the universe shouldn't make free-floating objects through direct collapse.

Except it does.

A new study using the European Space Agency's Euclid mission has pushed the census of free-floating planets in the σ Orionis cluster down to four Jupiter masses. The initial mass function—the distribution of object masses in a population—shows no sign of dropping off. No cutoff. No threshold. Just a smooth continuation into the planetary-mass regime.

The Cluster Laboratory

σ Orionis is a young open cluster embedded in the Orion B giant molecular cloud, about 400 parsecs from Earth. The cluster is roughly 3 million years old—newborn on cosmic timescales. Most of its members haven't yet settled onto the main sequence. They're still contracting, still hot from gravitational collapse, still surrounded in many cases by dusty disks.

The region is dominated by the σ Orionis star itself, a hot O-type star that floods the surrounding space with ultraviolet radiation. This creates a bright ionization front called the IC 434 H II region. The famous Horsehead Nebula points toward σ Orionis from the foreground.

The hot star has cleared out gas and dust in its immediate vicinity, creating a low-reddening window where faint objects can be detected without heavy extinction. This makes σ Orionis ideal for hunting the faintest cluster members—the brown dwarfs and free-floating planets that would be invisible in more obscured regions.

Surveys over the past two decades have identified progressively lower-mass objects in σ Orionis. Ground-based telescopes reached down to about 5-6 Jupiter masses in the deepest studies. But contamination from background galaxies becomes severe at these limits. Every faint red source could be a distant galaxy rather than a nearby brown dwarf.

Euclid changes the equation.

Space-Based Precision

Euclid launched in 2023 with a primary mission focused on cosmology—mapping the large-scale structure of the universe to constrain dark energy and dark matter. But its instruments deliver exquisite angular resolution and depth over wide areas, making it unexpectedly powerful for stellar astrophysics.

The VIS (visible) instrument provides optical imaging at 0.1 arcsecond resolution. The NISP (near-infrared spectrometer and photometer) delivers near-infrared imaging and slitless spectroscopy. Together, they reach limiting magnitudes of 26.2 in the optical and 24.5 in the near-infrared over the standard observing sequence.

For a cluster at 400 parsecs with an age of 3 million years, this translates to sensitivity down to objects with masses around 4 Jupiter masses according to evolutionary models.

The Early Release Observations program targeted five nearby star-forming regions to showcase Euclid's capabilities beyond cosmology. One pointing covered part of σ Orionis, including the Horsehead Nebula, the NGC 2023 embedded cluster, and a portion of the IC 434 H II region. Total area: 0.58 square degrees.

The observation took place on October 2, 2023, using the standard reference observing sequence—four dithered exposures in both VIS and NISP. Due to a dithering pattern issue during the science verification phase, some gaps remained in the final mosaic, and about 5% of the field was covered by only one image, making cosmic ray removal less efficient. But the data quality was validated as science-ready.

The Benchmark Strategy

Identifying genuine cluster members among thousands of detected sources requires stringent selection criteria. Background galaxies vastly outnumber brown dwarfs at faint magnitudes. Foreground stars with high proper motion can contaminate the sample. Reddening variations across the field complicate color-based selections.

The research team anchored their approach in seven previously known σ Orionis members with spectroscopic confirmation. These benchmarks span spectral types from late-M to mid-L—brown dwarfs with masses from roughly 50 down to 15 Jupiter masses.

All seven benchmarks appeared as clean point sources in the Euclid data. Their SPREAD_MODEL parameters—a star-galaxy classifier developed for the Dark Energy Survey and adapted for Euclid—clustered tightly around zero, confirming their point-source nature.

Two benchmarks showed slightly elevated FWHM values in the VIS images, suggesting they might be close binaries near the resolution limit. One of these, S Ori 52, revealed a faint companion at 0.96 arcsecond separation (387 AU projected) in the VIS image. The companion is 3.2 magnitudes fainter in the optical but only 4 magnitudes fainter in the near-infrared, giving it a bluer color than expected for a physical companion. Binary or chance alignment? The jury is out.

Finding two possible binaries among seven benchmarks is intriguing. Binary fractions among pre-main-sequence stars in Orion run around 12% for wide systems. Recent JWST observations of the Trapezium cluster suggested substellar binaries might be common. A dedicated HST survey of 33 brown dwarfs in young clusters found no resolved binaries wider than 20 AU, but that survey targeted older objects.

The point: even with just seven objects, Euclid's resolution is revealing new structure.

Filtering the Catalogue

The full Euclid catalogue for the σ Orionis field contained over 300,000 sources. The vast majority are background galaxies.

The team developed a high-purity selection procedure using the properties of the seven benchmarks. They calculated the mean and standard deviation of various parameters for the benchmarks, then defined cuts at ±1σ around the mean.

Selection criteria included: detection in the J-band (near-infrared sources are expected to dominate for young brown dwarfs), tight SPREAD_MODEL values in all bands (rejecting extended sources), restricted FWHM ranges (rejecting poorly resolved or saturated objects), and low ellipticity (rejecting edge-on galaxies and artifacts).

They divided the field into two regions based on reddening. About 70% of the field has visual extinction below 1.7 magnitudes—the low-reddening zone cleared by σ Orionis. The remaining 30% includes the Horsehead Nebula and NGC 2023, where extinction reaches much higher values.

The selection cuts were calibrated on the low-reddening region. Applying them reduced the sample from over 300,000 to about 3,300 objects—a 99% rejection rate.

These survivors were plotted on a color-magnitude diagram: optical magnitude versus optical-infrared color. The 3-million-year isochrone from the ATMO evolutionary models—recently validated using dynamical masses of brown dwarf binaries—was overlaid on the data.

The seven benchmarks defined a clear sequence parallel to the isochrone. Previously known photometric candidates from ground-based surveys fell along this sequence. And seven new objects appeared near the sequence, extending it to fainter magnitudes than any previous study.

Three of the new objects were bright enough to have counterparts in the VISTA Hemisphere Survey, a ground-based near-infrared catalogue. Positional agreement within 100 milliarcseconds confirmed they have low proper motions, consistent with cluster membership.

What the Data Show

The color-magnitude diagram tells a straightforward story. Objects follow the predicted cooling sequence for 3-million-year-old brown dwarfs and planets. The faintest new discoveries reach optical magnitudes of 26.4 and 26.5—right at Euclid's detection limit for point sources in this region.

At 403 parsecs with an age of 3 million years, the ATMO models predict that a 4-Jupiter-mass object should have the observed luminosity and colors. This marks Euclid's detection limit for this study.

But detection and completeness are different things. Number counts versus magnitude show that sources start becoming incomplete around magnitude 25 in the optical after all the filtering. This corresponds to about 6 Jupiter masses. Below that mass, the survey increasingly misses objects.

So: detections down to 4 Jupiter masses, completeness down to 6 Jupiter masses.

A color-color diagram using three near-infrared bands provides an additional check. The coolest benchmarks—spectral types L2 to L4.5—occupy a well-defined locus separated from the cloud of contaminating sources. The new candidates labeled D, E, and G fall squarely in this locus. Candidate F sits slightly to one side, potentially marking the transition from L to T spectral types where methane absorption begins to affect the colors.

Spectroscopy will be needed to confirm these classifications. But photometrically, they look like free-floating planets.

The Mass Function

Combining Euclid's substellar census with Gaia's stellar census produces a mass function spanning from about 0.15 solar masses down to 3 Jupiter masses—a dynamic range of 50 in mass, or five Jupiter masses up to 150 Jupiter masses.

The mass function is typically expressed as dN/dM, the number of objects per unit mass interval. Plot it on a log-log scale and the slope reveals the underlying physics of star formation.

For σ Orionis, the mass function follows different power laws in different mass regimes. From 0.15 to 0.1 solar masses (very low mass stars), the slope is 0.26±0.10. From 0.1 down to 0.011 solar masses (brown dwarfs), the slope is 0.18±0.01. From 0.011 down to 0.003 solar masses—the planetary-mass regime from 11 down to 3 Jupiter masses—the slope is 0.12±0.02.

These are shallow slopes. The mass function is close to flat across the substellar regime. There is no steepening at the low-mass end. No turnover. No cutoff.

Compare this to the field population—brown dwarfs in the solar neighborhood studied by comprehensive surveys out to 20 parsecs. A recent census found the field mass function steepens from a slope of 0.25 in the range 0.05-0.22 solar masses to 0.60 in the range 0.01-0.05 solar masses.

σ Orionis shows no such steepening. The slope stays flat or even gets slightly shallower at the planetary-mass end, though the error bars allow consistency with a constant slope.

What This Means

The smooth continuation of the mass function into the planetary-mass regime has implications for formation theory.

Opacity-limited fragmentation predicts a minimum mass around 0.01 solar masses (10 Jupiter masses) for objects forming through direct gravitational collapse. Below that mass, fragments can't cool efficiently enough to continue contracting. They should disperse or get accreted onto larger clumps before forming bound objects.

Yet the observations show free-floating objects down to 3-4 Jupiter masses with no sign of a cutoff at the detection limit.

Several mechanisms could explain planetary-mass objects in young clusters. Turbulent fragmentation in molecular clouds can produce smaller clumps than spherical collapse models predict. Gravitational fragmentation in filamentary structures might reach lower masses than in three-dimensional clouds. Ejection from protoplanetary disks could populate the field with planets that formed in disks but were dynamically scattered out. Photoevaporation by nearby massive stars could truncate cores and produce lower-mass objects.

The shape of the mass function provides clues. A multi-power-law distribution like the one observed in σ Orionis could arise from the interplay between mass-dependent and time-dependent growth in a population of accreting protostars. Different formation mechanisms might dominate in different mass regimes, producing the observed breaks in slope.

Or the mass function might be more universal than previously thought. If the same fragmentation and accretion processes operate from stellar masses down through brown dwarfs into the planetary regime, you'd expect a continuous distribution without sharp cutoffs.

The debate continues. What's clear is that free-floating planets are not rare edge cases. They're a natural extension of the stellar population in young clusters.

Complications and Caveats

Several systematic uncertainties affect the mass function determination. The quoted errors (±0.10, ±0.01, ±0.02 for the three regimes) reflect only the statistical uncertainty in fitting power laws to the observed distribution. They don't include model uncertainties.

Converting observed luminosities to masses requires evolutionary models. The ATMO models used here have been validated against dynamical mass measurements for brown dwarf binaries and perform better than older model sets. But uncertainties remain, particularly at the youngest ages and lowest masses where deuterium burning and atmospheric physics become important.

The cluster age is uncertain. A range of 3±2 million years allows a factor of several variation in inferred masses for a given luminosity. Older ages push masses lower. Younger ages push them higher.

Unresolved binaries complicate the picture. If 20% of objects are equal-mass binaries (plausible based on surveys of young brown dwarfs), each unresolved pair appears as a single object about 0.75 magnitudes brighter than the individual components. This shifts the inferred masses higher and steepens the mass function slopes.

The team ran simulations adding 20% equal-mass binaries randomly to the sample. The resulting binary-corrected mass function showed steeper slopes in all three regimes, though still shallower than the field population and still showing no cutoff.

Both photometric incompleteness and unresolved binaries work in the same direction: they cause the survey to underestimate the number of free-floating planets, particularly at the low-mass end below 6 Jupiter masses.

So the conclusion that there's no evidence for a cutoff is robust. If anything, the true mass function likely extends even more smoothly to lower masses than the observed one.

Looking Forward

This study covered 0.58 square degrees of one young cluster with one Euclid pointing. The Early Release Observations program includes four additional pointings of nearby star-forming regions: NGC 1333 in Perseus, Barnard 30, the Messier 78 dark clouds, and several Taurus molecular clouds.

Each region has different ages, densities, stellar populations, and environmental conditions. Comparing their substellar mass functions will reveal whether the σ Orionis result is universal or environment-dependent.

Euclid's main survey will cover 15,000 square degrees for cosmology. Parts of that footprint will include more distant young clusters and star-forming regions where Euclid's depth and resolution can extend existing catalogs. The mission lifetime allows for multi-epoch observations that will enable proper motion measurements to confirm cluster membership and measure velocity dispersions.

The Near-Infrared Spectrograph (NISP) provides slitless spectroscopy over the same fields as the imaging. Low-resolution spectra of thousands of brown dwarfs and planets will become available as the mission progresses, enabling spectroscopic confirmation of candidates and atmospheric characterization.

And this is just the beginning. The James Webb Space Telescope can follow up individual objects for high-resolution spectroscopy and resolved imaging of potential binaries. Ground-based adaptive optics systems on large telescopes can measure radial velocities and search for signatures of accretion or outflows.

The population of free-floating planets in young clusters is coming into focus. They're numerous. They're down there in the 3-5 Jupiter mass range. And they don't show the cutoff that classical theory predicted.

Whether they form like stars through direct fragmentation, or like planets in disks followed by ejection, or through some combination of processes, remains an open question. The mass function alone can't distinguish formation mechanisms conclusively.

But it can tell us this: the process that makes stars and brown dwarfs doesn't stop cleanly at some threshold mass. It continues smoothly into the regime of objects that have planetary masses but stellar birthplaces.

Orphan worlds. Nomads. Free-floating planets. Whatever we call them, they're out there in abundance, and Euclid is bringing them into view.

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/10.1051/0004-6361/202450793