Molecular Clouds Behave Differently in Varied Galaxy Environments

The Neighborhood That Shapes a Cloud

A molecular cloud is not a loner. For decades, astronomers treated these vast, cold nurseries of stars as if they lived in isolation, each one a self-contained factory where gravity slowly won the battle against turbulence and magnetic fields. You could study a cloud in the Milky Way, measure its mass, its size, its internal motions, and assume that similar clouds elsewhere would behave the same way. It was a reasonable assumption. It was also wrong.

In 2022, a team led by Jiayi Sun published what might be the most comprehensive census ever of molecular clouds across different galaxies. They used the PHANGS ALMA survey, which mapped carbon monoxide emission from 80 galaxies at resolutions fine enough to resolve individual clouds. What they found upends the simple picture: a molecular cloud's properties mass, size, internal velocity dispersion depend more on what is happening in its immediate galactic neighborhood than on anything intrinsic to the cloud itself. The environment writes the rules.

This is not a small revision. It means that star formation models built on local cloud data may fail when applied to other galaxies. It means that the galaxy itself, not just the cloud, is the relevant unit of analysis. And it raises a deeper question: if clouds are so sensitive to their surroundings, what does that say about the universality of star formation across the cosmos?

How Do You Survey 80 Galaxies Worth of Clouds?

The PHANGS project stands for Physics at High Angular Resolution in Nearby GalaxieS. It is a collaboration that combines the Atacama Large Millimeter/submillimeter Array (ALMA) with optical, ultraviolet, and infrared observations from the Hubble Space Telescope, the Very Large Telescope, and other facilities. The goal is to connect the small scale physics of star formation with the large scale properties of galaxies.

Sun and colleagues focused on the ALMA CO (2-1) survey, which maps the emission from carbon monoxide molecules. CO is the standard tracer of molecular hydrogen, the primary fuel for star formation. In cold molecular clouds, hydrogen is invisible directly, but CO glows brightly at millimeter wavelengths. By mapping CO, astronomers map where stars are born.

The team divided each of the 80 galaxies into kiloparsec scale apertures. A kiloparsec is about 3,260 light years, roughly the distance from the Sun to the edge of the Orion spiral arm. Within each aperture, they measured the surface densities of molecular gas, atomic gas, stars, and star formation rate. They also calculated orbital velocities and shear rates, which describe how fast material rotates around the galactic center and how much the rotation stretches or compresses the gas.

Then they identified individual molecular clouds within each aperture. They measured each cloud's mass, size, velocity dispersion, and virial parameter (a ratio that tells you whether a cloud is gravitationally bound or being torn apart by turbulence). Finally, they averaged these cloud properties across all the clouds in each aperture and asked: which environmental variables best predict those averages?

The sample is enormous: over 80 galaxies, thousands of kiloparsec scale regions, and tens of thousands of individual clouds. This is not a study of a few nearby clouds. It is a statistical census of star forming gas across the local universe.

The Two Variables That Matter Most

Sun and colleagues found that two environmental variables dominate the prediction of cloud properties: the surface density of molecular gas and the surface density of star formation rate. Both are measured at the kiloparsec scale, not at the cloud scale.

When molecular gas surface density is high, clouds tend to be more massive, larger, and have higher internal velocity dispersions. When star formation rate surface density is high, clouds also tend to be more massive and more turbulent. These correlations are strong. The authors ran a variable selection analysis to test which environmental properties carried the most predictive power. The molecular gas and star formation rate densities consistently came out on top.

Once those two variables are accounted for, the global properties of the host galaxy such as its total stellar mass, its morphological type, or its overall star formation rate added almost no additional predictive information. This is the key result: the apparent differences between cloud populations in different galaxies are not driven by something special about each galaxy. They are driven by local conditions that vary from one kiloparsec scale region to another.

An analogy: imagine comparing forests across continents. A forest in the Pacific Northwest looks different from a forest in the Amazon, but those differences are not because of something intrinsic to each continent. They are because of local conditions like rainfall, temperature, and soil nutrients. If you measure those local variables, the continent label becomes irrelevant. The same logic applies to molecular clouds.

Why This Changes the Picture

Before this study, many astronomers assumed that molecular clouds had relatively uniform properties across galaxies. The classic picture, developed from studies of Milky Way clouds, held that clouds have masses around 10,000 to 100,000 solar masses, sizes of tens of parsecs, and lifetimes of a few million years. Variations were attributed to differences in the cloud formation process itself.

Sun's results suggest that the variation is not random. It is systematically tied to the galactic environment. A cloud living in a dense, star forming region of a spiral arm will be different from a cloud in the diffuse outer disk of the same galaxy. And a galaxy with high average gas density will host a different population of clouds than a galaxy with low average gas density, simply because more of its area is filled with high density environments.

This has practical consequences. If you want to model star formation in a distant galaxy, you cannot assume the same cloud mass function or cloud lifetime that works for the Milky Way. You need to know the local gas density and star formation rate density in that galaxy. Those two numbers, once measured, will tell you what the cloud population looks like.

The study also provides a physical explanation for the Kennicutt Schmidt relation, which connects the surface density of gas to the surface density of star formation across galaxies. That relation has been known for decades, but its origin at the cloud scale was unclear. Sun's results suggest that the relation emerges because the properties of individual clouds, which directly control star formation, are themselves set by the same large scale gas density that appears in the Kennicutt Schmidt relation. It is a self consistent picture: the environment shapes the clouds, and the clouds shape the star formation rate.

Timescales: How Fast Does This All Happen?

The PHANGS dataset also allowed Sun and colleagues to estimate several important timescales. These numbers put the cloud evolution process into human terms.

The freefall time, which is the time it would take for a cloud to collapse under its own gravity if nothing stopped it, is about 5 to 20 million years. The turbulence crossing time, which is the time for sound waves to travel across a cloud, is similar. These numbers match previous estimates of cloud lifetimes. A typical molecular cloud lives for about 10 to 20 million years before it is disrupted by newly formed stars.

The orbital timescale, which is how long it takes for gas to circle the galaxy, is much longer: about 100 million years. The shear timescale, which measures how quickly differential rotation stretches a cloud, is also about 100 million years. Cloud cloud collision timescales are similarly long.

This means that clouds form and disperse much faster than the galaxy rotates. A cloud does not have time to complete a full orbit before it is destroyed. It also means that cloud cloud collisions, which some theories proposed as a major trigger of star formation, are too slow to be the primary mechanism. The clouds are born and die in less time than it takes for two of them to find each other.

The molecular gas depletion time, which is the time it would take to convert all the gas into stars at the current star formation rate, is 1 to 3 billion years. That is much longer than the cloud lifetime. Most of the gas in a galaxy is not in star forming clouds at any given moment. It sits in diffuse, non star forming gas, waiting to be assembled into clouds.

The authors found weak to no correlation between the depletion time and the other timescales. This is puzzling. It suggests that the efficiency of star formation, which is the fraction of gas that turns into stars per freefall time, is not simply set by the local dynamical conditions. Something else is regulating it, possibly feedback from supernovae or magnetic fields.

What This Research Does Not Prove

It is important to be clear about the limits of this study. The PHANGS survey covers 80 galaxies, but they are all relatively nearby, within about 50 million light years. These are normal, star forming spiral galaxies. The results may not apply to extreme environments like starburst galaxies, where gas densities are orders of magnitude higher, or to dwarf galaxies, where the low metallicity changes the chemistry of CO emission.

The study also relies on CO as a tracer of molecular gas. At low metallicities, CO becomes harder to detect because the carbon and oxygen abundances are lower. Some molecular gas may be present but invisible in CO. This is known as the "CO dark" gas problem. The PHANGS galaxies are mostly metal rich, so this is less of an issue, but it remains a caveat.

The analysis is correlational. The authors show that molecular gas density and star formation rate density predict cloud properties, but they do not prove causation. It could be that clouds with certain properties cause higher star formation rates, rather than the other way around. The timescale argument suggests that the environment acts first: the kiloparsec scale conditions are set by galactic dynamics, and clouds form within those conditions. But the direction of causality is not definitively settled.

Finally, the study treats clouds as discrete objects with well defined boundaries. In reality, molecular gas is a continuous, turbulent medium. The algorithm that identifies clouds imposes a threshold on the data, and different algorithms can produce different cloud catalogs. The authors used a standard approach, but the results may depend on the specific cloud identification method.

The Open Questions That Remain

If the environment controls cloud properties, what controls the environment? The kiloparsec scale gas density is set by galactic dynamics: spiral arms, bars, and differential rotation. But those structures are themselves shaped by the galaxy's history of mergers, accretion, and star formation feedback. The chain of causation extends from the cosmological scale down to the cloud scale, and this study only captures one link.

Another open question is whether the same relationships hold at higher redshifts, in galaxies that are forming stars much more rapidly. The PHANGS galaxies are local and relatively quiescent. Early universe galaxies have gas densities that are ten to a hundred times higher. If the scaling relations from this study extrapolate, those galaxies should have clouds that are much more massive and turbulent. But we do not yet have ALMA observations at sufficient resolution to test this directly.

A third question is about the role of magnetic fields. The study did not include magnetic field measurements because they are not available for most PHANGS galaxies. Magnetic fields are known to support clouds against collapse and to regulate star formation efficiency. It is possible that the residual scatter in the cloud property predictions, the variation not explained by gas density and star formation rate, is due to magnetic fields. Future surveys with the Square Kilometre Array or the Next Generation Very Large Array could test this.

What This Actually Means

• ▸Galaxy type matters less than neighborhood. If you want to predict what a molecular cloud looks like, do not ask what kind of galaxy it lives in. Ask about the local gas density and star formation rate in that specific region. Those two numbers carry almost all the predictive power.

• ▸Cloud lifetimes are short compared to galactic timescales. A cloud lives for 10 to 20 million years, while the galaxy takes 100 million years to rotate. This means clouds are transient structures that form and disperse within a single passage through a spiral arm. They do not persist long enough to be shaped by large scale shear or cloud cloud collisions.

• ▸Star formation efficiency is not set by local dynamics alone. The depletion time does not correlate with the other timescales. Something else, probably feedback from young stars or magnetic fields, regulates how much of the gas actually turns into stars. This is an active area of research.

• ▸The Kennicutt Schmidt relation has a cloud scale origin. The well known correlation between gas density and star formation rate across galaxies emerges because the same gas density that sets the star formation rate also sets the properties of the individual clouds that produce the stars. The relation is not a coincidence; it is built into the physics of cloud formation.

• ▸Future surveys need to measure environment, not just clouds. Studying molecular clouds in isolation, without measuring the kiloparsec scale conditions around them, will miss the most important variables. The PHANGS approach of combining high resolution cloud maps with lower resolution environmental maps should become the standard for extragalactic star formation studies.

Molecular clouds are not islands. They are products of their neighborhoods, shaped by the gas and stars around them. The galaxy is not just a backdrop; it is an active participant in the birth of every star.