Molecular Clouds and Gravitational Collapse

Stars form only in the coldest, densest phase of the interstellar medium, the giant molecular clouds — self-gravitating complexes of $10^5$ to $10^6{M}_{\odot }$ of molecular hydrogen, tens of parsecs across, with mean densities of $10^2$–$10^3{\text{cm}}^{-3}$ and temperatures near $10\text{K}$. At these low temperatures thermal pressure is feeble, and the question of whether a region collapses is a contest between self-gravity and the several forms of support: thermal pressure, turbulence, and magnetic fields. This lesson derives the threshold for collapse from the virial theorem, the timescale on which an unstable clump falls together, the fragmentation cascade that divides a cloud into many stars, and the magnetic physics that regulates the whole process.

Giant molecular clouds and their support

A molecular cloud is bound by its own gravity, but it does not collapse freely. The balance of forces is captured by the virial theorem, which for a cloud in equilibrium relates the internal kinetic (thermal plus turbulent) energy $K$, the gravitational potential energy $U$, and, in general, the magnetic and surface-pressure terms. In the simplest thermal-only form,

for a uniform sphere of mass $M$, radius $R$, and mean molecular weight $\mu$ (about $2.3$ for molecular gas with helium). Equilibrium is the boundary case: if the gravitational term dominates, $|U|>2K$, the cloud is unbound against its own gravity and collapses; if thermal energy dominates, it disperses or is confined only by external pressure. The collapse condition $|U|>2K$ is the physical content of the Jeans criterion.¹

Real clouds also carry supersonic turbulence and threading magnetic fields, both of which add support and postpone collapse. Including them, the full virial balance reads $2K+2K_{\text{turb}}+U+M_{\text{mag}}=0$, and observed giant molecular clouds sit close to virial equilibrium, neither collapsing wholesale nor flying apart. Only localized dense cores, where the balance tips, actually form stars.

The Jeans mass and Jeans length

Setting $|U|=2K$ at the boundary and solving for the mass gives the minimum mass a region of temperature $T$ and density $\rho$ must exceed to collapse. Eliminating the radius through $R=(3M/4\pi \rho {)}^{1/3}$ yields the Jeans mass,

and the corresponding length scale, the Jeans length, is the size below which a perturbation is stabilized by pressure and above which it collapses,

with $c_s$ the isothermal sound speed. The Jeans length is, up to a factor of order unity, the distance a sound wave travels in a free-fall time: a perturbation collapses only if gravity acts faster than pressure can respond across it.

The free-fall time

Once a clump exceeds the Jeans mass and pressure support becomes irrelevant, it collapses under gravity alone. The free-fall time is the time for a pressure-free, uniform sphere to collapse to a point. Each mass shell falls under the mass interior to it; integrating the equation of motion for a shell starting from rest gives

which depends only on the mean density, not on the size of the cloud — all shells arrive at the centre simultaneously for a uniform sphere. For the clump above, $t_{\text{ff}}\approx 10^6\text{yr}$, short compared to the million-to-ten-million year lifetimes of the clouds themselves. That collapse does not happen everywhere at this rate is direct evidence that additional support — turbulence and magnetic fields — slows the process; the observed star-formation rate of the Galaxy is roughly a hundred times below what unimpeded free-fall of all molecular gas would give.

The density dependence has a runaway character. As a region collapses its density rises, so $t_{\text{ff}}$ shrinks and the densest sub-regions collapse fastest. A uniform cloud is therefore unstable to fragmentation: small overdensities run ahead of the global collapse and separate out as independent collapsing centres.

The fragmentation cascade and the minimum mass

While the gas stays cold and optically thin, it collapses isothermally: the compressional heat is radiated away efficiently, so $T$ stays near $10\text{K}$ and the density climbs. Along an isothermal collapse the Jeans mass drops as $M_J\propto {\rho }^{-1/2}$, so a clump that was marginally unstable as a whole soon contains many new Jeans masses. Each fragments in turn, and the process repeats: a hierarchical fragmentation cascade subdivides the cloud into smaller and smaller bound pieces, which is why molecular clouds form gravitationally bound star clusters rather than single stars.²

The cascade cannot continue indefinitely. As fragments grow dense they become opaque to their own thermal radiation. When the collapse can no longer radiate away its compressional energy, it turns adiabatic: the temperature rises with further compression, $T\propto {\rho }^{\gamma -1}$, and now $M_J\propto {\rho }^{(3\gamma -4)/2}$ increases with density for $\gamma >4/3$. Fragmentation halts at the opacity limit, the density at which the gas first traps its own radiation. A short calculation of when the radiated luminosity of a collapsing fragment falls below its compressional heating rate gives a minimum fragment mass of order

a few Jupiter masses. This sets the low-mass end of star formation: objects below the opacity-limited minimum cannot form by direct fragmentation, and the smallest stellar fragments are near the boundary between the lowest-mass stars and brown dwarfs.

The initial mass function

The distribution of masses with which stars are born, the initial mass function (IMF), is the statistical outcome of the fragmentation cascade. Empirically it is a declining power law in mass: many low-mass stars form for every high-mass one. Salpeter first fit the number of stars per unit mass as

a slope that holds well above half a solar mass. At lower masses the function flattens; the Kroupa parametrization uses a broken power law, with slope $-2.3$ above $0.5M_{\odot }$, $-1.3$ between $0.08$ and $0.5M_{\odot }$, and a further flattening into the brown-dwarf regime below the hydrogen-burning limit at $0.08M_{\odot }$. The turnover near a few tenths of a solar mass marks the characteristic mass of star formation, plausibly tied to the Jeans mass at the density where cloud cores become opaque.

Magnetic support and ambipolar diffusion

Molecular clouds are threaded by magnetic fields of $10$–$100\mu \text{G}$, and the field resists compression across its lines. The magnetic support of a cloud is set by its mass-to-flux ratio $M/\Phi$, where $\Phi =B\pi R^2$ is the magnetic flux through it. There is a critical value

below which the magnetic tension can support any compression and the cloud is subcritical (stable against collapse), and above which it is supercritical and collapses regardless of field strength. Because flux is conserved as gas is compressed along field lines, the mass-to-flux ratio is nearly invariant during collapse, so a subcritical cloud stays subcritical.

The way a subcritical cloud forms stars is ambipolar diffusion. The magnetic field is tied only to the ionized component; the neutral gas, which is the overwhelming majority, feels the field only indirectly through ion–neutral collisions. In the weakly ionized interior of a molecular cloud (ionization fraction $\sim 10^{-7}$, maintained by cosmic rays), the neutrals slowly drift inward through the ions and the field, concentrating mass at the centre while leaving the flux behind. Over a characteristic time of $\sim 10\text{Myr}$ the central region becomes supercritical and collapses dynamically. Ambipolar diffusion is the slow, rate-limiting step that makes the observed star-formation efficiency low even though the free-fall time is short.

The Jeans mass and free-fall time fix when and how fast a cloud collapses, the fragmentation cascade and its opacity-limited cutoff fix the range of masses produced, and ambipolar diffusion sets the slow clock in magnetically supported clouds. What happens to a single collapsing core once it becomes optically thick — the birth of a protostar and its climb toward the main sequence — is the subject of the next lesson.