Protostellar Collapse and the Class 0–III Sequence

A star begins as a failure of pressure. In the densest, coldest pockets of a molecular cloud — number density $n\gtrsim10^4\,\mathrm{cm^{-3}}$ , temperature $T\sim10\,\mathrm{K}$ — self-gravity overwhelms the thermal support and a prestellar core starts to fall inward on itself. The early collapse is nearly isothermal, because the released gravitational energy escapes freely as infrared photons from dust: the density climbs through many decades while the temperature barely stirs. Only when the centre grows opaque to its own dust emission does the trapped heat stiffen the gas into a first hydrostatic core, and a second collapse — triggered by $\mathrm{H_2}$ dissociation — plants the true stellar embryo at its heart. From that moment the object is a protostar buried inside an infalling envelope, and we never see it directly. What reaches the telescope is reprocessed light: stellar luminosity absorbed by dust and re-emitted from far-infrared to millimetre. So astronomers classify young stellar objects not by their hidden interiors but by the shape of their spectral energy distribution. The Class 0 / I / II / III scheme — built from the spectral index $\alpha_{\mathrm{IR}}$ and the bolometric temperature $T_{\mathrm{bol}}$ — orders sources from envelope-swaddled to photosphere-bare, and reads as an approximate evolutionary clock. Underneath the taxonomy runs one storyline: the disposal of mass and angular momentum. Infall builds the star and, because a rotating core cannot fall straight in, a centrifugally supported disk; bipolar jets carve polar cavities and shed spin; and accretion, photoevaporation, and planet formation disperse the disk over a few Myr, until a Class III pre-main-sequence star emerges to descend its Hayashi track toward the main sequence.

Key takeaways

Star formation starts when a dense cloud core exceeds its Jeans mass and collapses in near free fall, $t_{\mathrm{ff}}\sim(G\rho)^{-1/2}$ , forming a protostar inside an infalling envelope.
Because the protostar is buried, the Class 0–I–II–III sequence orders young stellar objects by the slope and peak of their infrared-to-submillimetre SED, tracking the dispersal of envelope and disk.
Class 0 are deeply embedded with $M_{\mathrm{env}}>M_\star$ ; Class I retain a residual envelope plus disk; Class II are optically visible T Tauri stars with disks; Class III show a nearly bare photosphere.
The two discriminants are the near-IR spectral index $\alpha_{\mathrm{IR}}=\mathrm{d}\log(\lambda F_\lambda)/\mathrm{d}\log\lambda$ and the bolometric temperature $T_{\mathrm{bol}}\propto\langle\nu\rangle$ .
Angular-momentum conservation forces a $\sim10^2\,\mathrm{AU}$ disk; the steady Shu rate $\dot M\approx0.975\,a^3/G\sim10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$ feeds the star while outflows shed spin.
Durations are $\sim10^4\,\mathrm{yr}$ (Class 0), $\sim10^5\,\mathrm{yr}$ (Class I), and a few $\times10^6\,\mathrm{yr}$ (Class II), set by free fall, envelope dispersal, and disk dissipation.
Key idea: the SED-defined Class 0–III sequence is a proxy for the physical progression of star formation — near-free-fall collapse builds an embedded protostar, angular-momentum conservation forms a disk, outflows and accretion clear the envelope, and disk dissipation finally exposes a pre-main-sequence photosphere.

The core insight

The observational Class 0–III sequence, defined by SED shape ( $\alpha_{\mathrm{IR}}$ , $T_{\mathrm{bol}}$ ), is a proxy for the physical progression of star formation: near-free-fall collapse builds an embedded protostar, angular-momentum conservation forms an accretion disk, outflows and accretion disperse the envelope, and disk dissipation finally exposes a pre-main-sequence photosphere.

Where this lives today

The Class 0–III framework is the working language of modern star-formation surveys. Space telescopes — Spitzer, Herschel, and now JWST — build infrared-to-submillimetre SEDs of thousands of young stellar objects, measuring $\alpha_{\mathrm{IR}}$ and $T_{\mathrm{bol}}$ to sort populations and time each phase. On the ground, ALMA resolves cold dust and gas at tens-of-AU scales, imaging Class 0 envelopes with their disks and outflows and weighing $M_{\mathrm{env}}$ from the optically thin millimetre continuum. Its ringed, gapped Class II disks (the DSHARP survey) show planet formation already under way as the disk dissipates, tying the end of the sequence to the birth of planetary systems. The luminosity problem stays active: variability monitoring and FU Orionis outburst studies test how episodic accretion really is, while candidate first hydrostatic cores are hunted among very low-luminosity objects — closing the last observational gap.

The physical picture

Picture squeezing a cold sponge of gas. At first it compresses almost for free, because every scrap of released gravitational energy leaks away as infrared light from dust grains, holding the temperature near 10 K while the density races upward. Only when the interior becomes opaque does the trapped heat build enough pressure to push back, and a hydrostatic core — the seed of a star — condenses out. This core is hidden; the object announces itself only through the cocoon of dust that soaks up the buried luminosity and re-radiates it in the far-infrared and millimetre. As the cocoon thins, the peak of the light marches from the millimetre (cold envelope) toward the near-infrared and optical (warm photosphere) — the bluewards march the Class 0 → III numbers measure. The first figure follows one core through the four stages, pinning to each the two observables: how long it lasts and how cold its bolometric temperature is.

Figure

A prestellar core through the four classes — the amber envelope disperses while a blue disk and red bipolar outflow appear, and the clock runs from t ∼ 10⁴ yr with T_bol ≲ 70 K in Class 0 to a few × 10⁶ yr and T_bol > 2800 K by Class III

Angular momentum is the antagonist of the whole story. A core rotating even slowly, $\Omega_0\sim10^{-14}\,\mathrm{s^{-1}}$ , carries far too much spin for its gas to fall straight onto a star a few solar radii across; the infall piles into a flattened disk at a centrifugal radius of order $10^2\,\mathrm{AU}$ . Growing the star means transporting angular momentum outward through the disk while mass moves in, and launching some material back out along the poles as a jet. The observable consequence is the second figure: the SED, whose slope $\alpha_{\mathrm{IR}}$ and bolometric temperature $T_{\mathrm{bol}}$ shift as envelope and disk are consumed. A cold single far-IR bump is Class 0; a rising near-to-far-IR SED is Class I; a falling SED with infrared excess is Class II; a nearly bare reddened photosphere is Class III.

Figure

Schematic SEDs λF_λ versus wavelength for the four classes, with the discriminants marked: Class I has α_IR greater than 0.3 measured between 2.2 and 20 μm, Class II lies between −1.6 and −0.3, Class III below −1.6, and Class 0 is set instead by a cold envelope with T_bol below 70 K

The core equations

The physical clock of collapse is the free-fall time — the interval for a cold, pressureless uniform sphere to collapse to a point. It depends only on the mean density, not the sphere's size — the signature of homologous collapse.

t_{\mathrm{ff}}=\sqrt{\dfrac{3\pi}{32\,G\,\rho_0}} \qquad (1)

where: $t_{\mathrm{ff}}$ — free-fall timescale; $G$ — gravitational constant; $\rho_0$ — initial mean mass density. Reading: denser gas collapses faster, and every shell arrives at the centre together. Collapse begins only if the core outweighs what pressure can support. The Jeans mass is that threshold, from balancing thermal energy against gravitational binding for a uniform sphere.

M_J=\left(\dfrac{5k_BT}{G\mu m_H}\right)^{3/2}\left(\dfrac{3}{4\pi\rho_0}\right)^{1/2} \qquad (2)

where: $M_J$ — minimum mass for instability; $k_B$ — Boltzmann constant; $T$ — gas temperature; $\mu$ — mean molecular weight ( $\approx2.3$ ); $m_H$ — hydrogen mass; $\rho_0$ — density. Reading: cold, dense gas has a small $M_J$ , so cooling or compression promotes collapse. Because the protostar is invisible, the working diagnostic is the slope of the SED plotted as $\lambda F_\lambda$ against $\lambda$ , measured across the near-to-mid infrared.

\alpha_{\mathrm{IR}}=\dfrac{\mathrm{d}\log(\lambda F_\lambda)}{\mathrm{d}\log\lambda}\Bigg|_{2.2\,\mu\mathrm{m}}^{\sim20\,\mu\mathrm{m}} \qquad (3)

where: $\alpha_{\mathrm{IR}}$ — SED slope in $\lambda F_\lambda$ versus $\lambda$ ; $F_\lambda$ — flux per unit wavelength; $\lambda$ — evaluated between $\sim2$ and $\sim20\,\mu\mathrm{m}$ . Reading: rising ( $\alpha_{\mathrm{IR}}>0$ ) means embedded, falling ( $<0$ ) means a photosphere shows through. The second discriminant compresses the whole SED into the temperature of a blackbody with the same flux-weighted mean frequency $\langle\nu\rangle$ — a thermometer of how buried the source is.

T_{\mathrm{bol}}=1.25\times10^{-11}\,\langle\nu\rangle\ \mathrm{K},\qquad \langle\nu\rangle=\dfrac{\int \nu F_\nu\,\mathrm{d}\nu}{\int F_\nu\,\mathrm{d}\nu} \qquad (4)

where: $T_{\mathrm{bol}}$ — bolometric temperature; $\langle\nu\rangle$ — flux-weighted mean frequency; $F_\nu$ — flux per unit frequency. Reading: the coefficient is $h/(3.83\,k_B)$ , so a true blackbody returns $T_{\mathrm{bol}}=T$ ; cold far-IR gives $\lesssim70\,\mathrm{K}$ (Class 0), a bare photosphere $\gtrsim2800\,\mathrm{K}$ (Class III). Once the singular isothermal sphere begins its inside-out collapse, the self-similar solution predicts a constant mass rate set only by the sound speed and $G$ .

\dot{M}=\dfrac{m_0\,a^3}{G}\approx0.975\,\dfrac{a^3}{G},\qquad a=\sqrt{\dfrac{k_BT}{\mu m_H}} \qquad (5)

where: $\dot M$ — accretion rate onto protostar plus disk; $a$ — isothermal sound speed; $m_0\approx0.975$ — similarity constant (Shu 1977). Reading: for $T=10\,\mathrm{K}$ , $a\approx0.19\,\mathrm{km\,s^{-1}}$ and $\dot M\sim10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$ ; hotter cores accrete faster. The gravitational energy liberated as infalling matter strikes the stellar surface is the accretion luminosity, which dominates an embedded protostar's energy budget.

L_{\mathrm{acc}}=\dfrac{G M_\star \dot{M}}{R_\star} \qquad (6)

where: $L_{\mathrm{acc}}$ — luminosity at the accretion shock; $M_\star,R_\star$ — protostellar mass and radius; $\dot M$ — accretion rate. Reading: a half-solar protostar at the Shu rate gives $L_{\mathrm{acc}}\sim10\,L_\odot$ , more than most embedded sources radiate — the luminosity problem. A rotating core cannot deposit its gas on the star: specific angular momentum is conserved, so material settles where centrifugal support equals gravity — the disk radius.

R_{\mathrm{disk}}\sim\dfrac{j^2}{GM_\star}=\dfrac{(\Omega_0 R_0^2)^2}{GM_\star} \qquad (7)

where: $R_{\mathrm{disk}}$ — centrifugal radius; $j$ — specific angular momentum; $\Omega_0$ — initial angular velocity; $R_0$ — initial shell radius; $M_\star$ — central mass. Reading: typical parameters give $R_{\mathrm{disk}}\sim10^2\,\mathrm{AU}$ , thousands of times $R_\star$ , so a disk is essentially guaranteed. The cold envelope itself is weighed through its modified-blackbody dust emission, $F_\nu^{\mathrm{env}}\propto\nu^{2+\beta}B_\nu(T_d)$ in the Rayleigh–Jeans regime (emissivity index $\beta\approx1$ – $2$ , $T_d\sim10$ – $30\,\mathrm{K}$ ): because the millimetre dust is optically thin, its flux calibrates $M_{\mathrm{env}}$ , the quantity that defines Class 0 through $M_{\mathrm{env}}>M_\star$ .

Derivation AThe free-fall time and why it ignores the size of the cloud

The goal is to derive the collapse timescale of a cold, self-gravitating sphere and to show that it depends only on the mean density, not on the radius — the fact that makes uniform collapse homologous, so a uniform core stays uniform down to the moment of runaway. We treat each shell as a test particle falling under only the interior mass, integrate once for its speed, and reduce the collapse integral with a trigonometric substitution.

Step 1. Neglecting pressure, Newton's second law for a shell at radius $r$ started from $r_0$ reads $\ddot r=-GM/r^2$ , with enclosed mass $M=\tfrac{4}{3}\pi r_0^3\rho_0$ held fixed. Why: no shell overtakes another, so the interior mass is constant — a one-body fall in a fixed central field.

Step 2. Multiply by $\dot r$ and integrate with $\dot r(r_0)=0$ for the energy integral $\tfrac12\dot r^2=GM\left(\tfrac1r-\tfrac1{r_0}\right)$ . Why: conservation of energy, giving the infall speed at any radius.

Step 3. Substitute $r=r_0\cos^2\theta$ (start $\theta=0$ , full collapse $\theta=\pi/2$ ); the time integral becomes $\int\cos^2\theta\,\mathrm{d}\theta$ , yielding $t_{\mathrm{ff}}=\dfrac{\pi}{2}\sqrt{\dfrac{r_0^3}{2GM}}$ . Why: the cycloid substitution linearizes the singular $1/r$ potential and delivers a finite time despite the diverging acceleration.

Step 4. Replace $M=\tfrac43\pi r_0^3\rho_0$ ; every power of $r_0$ cancels, leaving $t_{\mathrm{ff}}=\sqrt{3\pi/32G\rho_0}$ . For $n=10^4\,\mathrm{cm^{-3}}$ , $\rho_0\approx3.8\times10^{-20}\,\mathrm{g\,cm^{-3}}$ and $t_{\mathrm{ff}}\approx3.4\times10^5\,\mathrm{yr}$ . Why: since all shells share $t_{\mathrm{ff}}$ , they reach the centre together — the collapse stays uniform and accelerates as a whole. *Numerically: $t_{\mathrm{ff}}=\sqrt{3\pi/(32\times6.67\times10^{-8}\times3.85\times10^{-20})}\approx1.07\times10^{13}\,\mathrm{s}\approx3.4\times10^5\,\mathrm{yr}$ ; denser Class 0 cores ( $n\sim10^6$ ) give $\sim3\times10^4\,\mathrm{yr}$ .*

Go deeper · Derivation B · 1 minThe constant accretion rate of inside-out collapse

Derivation BThe constant accretion rate of inside-out collapse

The aim is to show why the self-similar collapse of a singular isothermal sphere delivers mass to the centre at a nearly constant rate $\dot M\approx0.975\,a^3/G$ , fixed by the sound speed alone, and to evaluate it for a cold core. We argue dimensionally that only one rate can be built from the two constants, recall how the similarity solution fixes the number, and trace the mechanism — an outward rarefaction wave peeling the cloud inside-out.

Step 1. The static singular isothermal sphere has density $\rho(r)=a^2/(2\pi G r^2)$ , a scale-free profile whose only parameters are $a=\sqrt{k_BT/\mu m_H}$ and $G$ . Why: scale-free initial conditions admit a self-similar solution — the collapse looks the same at every time once rescaled, so a single constant rate should emerge.

Step 2. From $a$ ( $\mathrm{cm\,s^{-1}}$ ) and $G$ ( $\mathrm{cm^3\,g^{-1}\,s^{-2}}$ ) the only mass-per-time combination is $a^3/G$ ; the Shu (1977) similarity solution fixes the coefficient to $m_0\approx0.975$ . Why: dimensions guarantee the form; only the exact solution supplies the order-unity number.

Step 3. At core formation a rarefaction wave launches from the centre outward at $a$ ; interior gas, its pressure support gone, falls in — the cloud collapses inside-out at fixed $\dot M$ . Why: the collapse front, not gravity on the whole cloud at once, sets the pace, which is why the rate is constant.

Step 4. For $T=10\,\mathrm{K}$ , $\mu=2.3$ : $a\approx0.19\,\mathrm{km\,s^{-1}}$ , so $\dot M\approx0.975\,a^3/G\approx1.0\times10^{20}\,\mathrm{g\,s^{-1}}\approx1.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$ . Why: this assembles a solar mass in a few $\times10^5\,\mathrm{yr}$ , matching the embedded Class 0 plus I lifetime and setting the accretion-luminosity scale. *Numerically: $a^3/G=(1.9\times10^4)^3/(6.67\times10^{-8})=1.03\times10^{20}\,\mathrm{g\,s^{-1}}$ ; $\times0.975\times(3.16\times10^7/1.99\times10^{33})=1.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$ .*

Go deeper · Derivation C · 1 minAngular-momentum conservation forces a disk

Derivation CAngular-momentum conservation forces a disk

The purpose is to demonstrate that a molecular-cloud core, rotating even at the slow rates actually observed, cannot deliver its gas directly onto the tiny protostar, because conservation of specific angular momentum halts infall at a centrifugal radius that dwarfs the stellar radius — so a disk is inevitable and accretion must proceed through it. We compute the specific angular momentum of an outer shell, set centrifugal support equal to gravity, and evaluate for realistic numbers.

Step 1. For solid-body rotation at $\Omega_0$ , a shell at initial radius $R_0$ carries $j=\Omega_0 R_0^2$ , conserved during collapse because central gravity exerts no torque. Why: with no torque the gas spins up as $1/R^2$ — the skater pulling in her arms.

Step 2. Infall stops where centrifugal acceleration $j^2/R^3$ balances gravity $GM_\star/R^2$ , giving $R_{\mathrm{disk}}=j^2/GM_\star=\Omega_0^2R_0^4/GM_\star$ . Why: inside this radius rotation wins, so gas settles into a rotationally supported ring — the seed of the disk.

Step 3. For $R_0\sim0.05\,\mathrm{pc}$ , $\Omega_0\sim10^{-14}\,\mathrm{s^{-1}}$ , $M_\star\sim0.5\,M_\odot$ : $j\approx2.4\times10^{20}\,\mathrm{cm^2\,s^{-1}}$ and $R_{\mathrm{disk}}\approx60\,\mathrm{AU}$ , versus $R_\star\sim10^{-2}\,\mathrm{AU}$ . Why: four orders of magnitude between $R_{\mathrm{disk}}$ and $R_\star$ mean nearly all the gas lands in a disk, matching observed 50–500 AU Class II disks.

Step 4. Growing the star then requires transporting angular momentum outward — viscous and magnetorotational stresses, gravitational torques — while outflows carry off the excess. Why: without a sink for angular momentum the disk would spin without accreting; the disk-plus-outflow engine drives Class 0 and I evolution. *Numerically: $j=10^{-14}\times(0.05\times3.09\times10^{18})^2=2.4\times10^{20}\,\mathrm{cm^2\,s^{-1}}$ ; $R_{\mathrm{disk}}=5.7\times10^{40}/(6.67\times10^{-8}\times9.95\times10^{32})=8.5\times10^{14}\,\mathrm{cm}=57\,\mathrm{AU}$ .*

Limiting cases

The dominant physics — and hence the observational signature — changes character across the collapse and along the class sequence:

Case	Result	Meaning
Isothermal early collapse	$P\propto\rho,\ T\approx10\,\mathrm{K}$	Optically thin dust radiates heat away as fast as compression makes it; density climbs many decades at fixed temperature.
Adiabatic first core	$P\propto\rho^{\gamma},\ \gamma\approx5/3$	Once opaque, compression heats the gas; a quasi-hydrostatic first core (a few AU) briefly halts collapse.
Free-fall dominated (Class 0)	$t_{\mathrm{ff}}=\sqrt{3\pi/32G\rho}$	Envelope infall dominates, $M_{\mathrm{env}}>M_\star$ , $T_{\mathrm{bol}}\lesssim70\,\mathrm{K}$ ; short-lived, $\sim10^4\,\mathrm{yr}$ .
Disk dominated (Class II)	$\alpha_{\mathrm{IR}}<0,\ 650\lesssim T_{\mathrm{bol}}\lesssim2800\,\mathrm{K}$	Envelope gone; IR excess from a disk atop a visible photosphere — a classical T Tauri star.
Bare photosphere (Class III)	$\lambda F_\lambda\propto\lambda^{-4},\ \alpha_{\mathrm{IR}}<-1.6$	Disk dissipated; SED approaches a reddened blackbody descending pre-main-sequence tracks.

Common trapThe classes are SED shapes, not guaranteed physical stages

The Class 0–III scheme is defined by the appearance of the SED, not directly by the interior state — and appearance depends on geometry. Real envelopes and disks are flattened and clumpy, so viewing angle matters: an edge-on Class II disk can hide the photosphere and mimic a far more embedded source, biasing $T_{\mathrm{bol}}$ toward younger-looking values, while a pole-on object seen down an outflow cavity can look more evolved than it is. Treat the boundaries as useful bins, not sharp thresholds, and never conflate the transient first and second hydrostatic cores (dynamical, years to millennia) with the observational Class 0 stage (statistical, $\sim10^4\,\mathrm{yr}$ ). Independent tracers — outflow force, envelope mass, kinematics — pin down the true state.

Worked examples

1 · Free-fall time of a prestellar core (fully worked)

A prestellar core has molecular-gas density $n=10^4\,\mathrm{cm^{-3}}$ ( $\mu=2.3$ ). Find its free-fall time, then repeat for a dense Class 0 core at $n=10^6\,\mathrm{cm^{-3}}$ .

Step 1. Convert density to mass density: $\rho_0=\mu m_H n=2.3\times(1.67\times10^{-24}\,\mathrm{g})\times10^4\,\mathrm{cm^{-3}}=\mathbf{3.85\times10^{-20}\,g\,cm^{-3}}$ .

Step 2. Insert into equation (1): $t_{\mathrm{ff}}=\sqrt{9.42/(32\times6.67\times10^{-8}\times3.85\times10^{-20})}=1.07\times10^{13}\,\mathrm{s}$ . Dividing by $3.16\times10^7\,\mathrm{s\,yr^{-1}}$ gives $t_{\mathrm{ff}}\approx\mathbf{3.4\times10^{5}\,yr}$ .

Step 3. For $n=10^6\,\mathrm{cm^{-3}}$ the density is $100\times$ higher, and since $t_{\mathrm{ff}}\propto\rho^{-1/2}$ the time falls by $\sqrt{100}=10$ : $t_{\mathrm{ff}}\approx\mathbf{3.4\times10^{4}\,yr}$ . This is why the deeply embedded Class 0 phase, drawing on the densest gas, lasts only $\sim10^4\,\mathrm{yr}$ — the rarest, briefest, hardest-to-catch snapshot in the sequence. A ten-fold density jump collapses ten times faster.

2 · Accretion luminosity and the luminosity problem (fully worked)

A protostar with $M_\star=0.5\,M_\odot$ , $R_\star=2.5\,R_\odot$ accretes at the Shu rate $\dot M=1.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$ . Find $L_{\mathrm{acc}}$ and compare with the $\sim1.5\,L_\odot$ median of observed embedded protostars.

Step 1. In cgs: $M_\star=9.95\times10^{32}\,\mathrm{g}$ , $R_\star=1.74\times10^{11}\,\mathrm{cm}$ , $\dot M=\mathbf{1.0\times10^{20}\,g\,s^{-1}}$ .

Step 2. Apply equation (6): $L_{\mathrm{acc}}=GM_\star\dot M/R_\star=(6.67\times10^{-8})(9.95\times10^{32})(1.0\times10^{20})/(1.74\times10^{11})=3.8\times10^{34}\,\mathrm{erg\,s^{-1}}$ .

Step 3. In solar units, $L_{\mathrm{acc}}=3.8\times10^{34}/3.83\times10^{33}\approx\mathbf{10\,L_\odot}$ — roughly seven times the observed embedded median.

Step 4. The mismatch is the luminosity problem. Its standard resolution is that accretion is episodic: gas piles up in the disk during long, dim, quiescent intervals and is dumped onto the star in short, bright bursts — the FU Orionis phenomenon. The time-averaged luminosity stays low even though the mass is assembled. Predicted $\sim10\,L_\odot$ far exceeds the observed $\sim1.5\,L_\odot$ — accretion cannot be steady.

Physics of the portrait

Protostellar collapse is where a century of stellar theory meets half a century of infrared observation: a process no one can watch directly, reconstructed from a cold spectrum and a few conserved laws.

Origins · 1 minprotostellar collapse and how this was discovered

protostellar collapse

the Class 0 to III young-stellar-object sequence · low-mass star formation

The idea that a star condenses from diffuse gas is old, but turning it into physics took the twentieth century. James Jeans in 1902 fixed the mass scale at which self-gravity beats pressure — the criterion a collapsing core must satisfy. Chushiro Hayashi in 1961 showed that a newly revealed pre-main-sequence star is fully convective and descends a nearly vertical Hertzsprung–Russell track, anchoring the endpoint. The decisive theoretical step came from Richard Larson in 1969, whose numerical collapse revealed the two-stage structure still taught today: isothermal contraction, a first hydrostatic core, a second collapse when $\mathrm{H_2}$ dissociates, and finally the stellar embryo. Frank Shu in 1977 supplied the analytic counterpart — the self-similar inside-out collapse of a singular isothermal sphere with its clean rate $\dot M\approx0.975\,a^3/G$ — developed with Fred Adams and Susana Lizano into the standard infall-disk-outflow picture. The observational taxonomy followed: Charles Lada in 1987 ordered young stellar objects by infrared spectral index into Classes I, II, and III, and André, Ward-Thompson, and Barsony in 1993 added Class 0 for the embedded, submillimetre-bright protostars whose envelopes outweigh their stars. The modern synthesis — an SED read as a clock — completes what Jeans began.

History

1902

— Jeans sets the mass scale at which self-gravity overcomes gas pressure

1969

— Larson's numerical collapse predicts the first and second hydrostatic cores

1977

— Shu solves the self-similar inside-out collapse of a singular isothermal sphere

1987

— Lada orders young stellar objects by infrared spectral index into Classes I to III

1993

— André, Ward-Thompson and Barsony define Class 0 from submillimetre envelopes

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 12 (Star Formation) · Stahler and Palla, The Formation of Stars (the standard monograph) · Shu, Adams and Lizano 1987, Ann. Rev. Astron. Astrophys. 25, 23 (inside-out collapse) · Larson 1969, MNRAS 145, 271 (numerical protostellar collapse) · Lada 1987, IAU Symp. 115, 1 (SED classification) · André, Ward-Thompson and Barsony 1993, ApJ 406, 122 (Class 0 definition) · Binney and Tremaine, Galactic Dynamics, 2nd ed. (gravitational collapse and virial arguments). The key anchor numbers are: $t_{\mathrm{ff}}\approx3.4\times10^5\,\mathrm{yr}$ for $n=10^4\,\mathrm{cm^{-3}}$ ; $\dot M\approx1.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$ for $T=10\,\mathrm{K}$ ; $L_{\mathrm{acc}}\approx10\,L_\odot$ for the worked-example parameters; and $R_{\mathrm{disk}}\approx60\,\mathrm{AU}$ . The class boundaries in $\alpha_{\mathrm{IR}}$ and $T_{\mathrm{bol}}$ follow the Greene, Chen, and André conventions; the coefficient $1.25\times10^{-11}\,\mathrm{K\,Hz^{-1}}$ in $T_{\mathrm{bol}}$ equals $h/(3.83\,k_B)$ .

Prerequisites. The Jeans Instability and Jeans Mass · Molecular Clouds and Giant Molecular Clouds · Virial Theorem for Molecular Clouds · Thermal Dust Emission and Infrared Radiation · Interstellar Extinction and Reddening. Leads to. Pre-Main-Sequence Evolution and Hayashi Tracks · Protoplanetary Disks and Planet Formation · Bipolar Outflows and Herbig–Haro Objects.

Ask the tutor: Ask me to walk through why a rotating cloud core cannot help but build a disk, or to classify a real young stellar object from its measured $\alpha_{\mathrm{IR}}$ and $T_{\mathrm{bol}}$ .

Why is the early collapse isothermal but the later collapse adiabatic?

It is a race between how fast compression heats the gas and how fast that heat escapes. Early on the core is optically thin to its own dust emission, so infrared photons stream out and carry off the heat almost instantly, holding $T\approx10\,\mathrm{K}$ while the density climbs through many decades. Once the centre becomes optically thick, radiation can no longer escape promptly; the trapped heat raises temperature and pressure, the equation of state stiffens toward $\gamma\approx5/3$ , and a hydrostatic first core briefly arrests the collapse. Isothermal versus adiabatic is simply transparent versus opaque.

How can Class 0 objects be identified if they are too embedded to see?

Precisely because they are embedded. A Class 0 protostar hides its star behind a massive cold envelope, but that envelope is optically thin at submillimetre wavelengths, so its dust glows as a modified blackbody, $F_\nu\propto\nu^{2+\beta}B_\nu(T_d)$ , and is directly detectable. Two signatures define the class: a low bolometric temperature ( $T_{\mathrm{bol}}\lesssim70\,\mathrm{K}$ ), meaning nearly all light emerges cold and far-IR, and a large submillimetre-to-bolometric ratio ( $L_{\mathrm{submm}}/L_{\mathrm{bol}}>0.5\%$ ), meaning $M_{\mathrm{env}}>M_\star$ . The optically thin flux even weighs the envelope directly.

Why must a disk form instead of the gas falling straight onto the star?

Because angular momentum is conserved and the star is minuscule. A core rotating at even $\Omega_0\sim10^{-14}\,\mathrm{s^{-1}}$ gives its outer gas a specific angular momentum $j=\Omega_0R_0^2$ it keeps as it collapses, since gravity exerts no torque. Infall halts where centrifugal support balances gravity, at $R_{\mathrm{disk}}=j^2/GM_\star\sim60$ – $100\,\mathrm{AU}$ — thousands of times a few- $R_\odot$ star — so the gas cannot reach the star and settles into a rotationally supported disk, through which it must then accrete while outflows carry off the excess spin.

Is the Class sequence really an evolutionary track in time?

Roughly, but with caveats. Statistically the classes are time-ordered — Class 0 rarest and youngest ( $\sim10^4\,\mathrm{yr}$ ), Class II longest-lived (a few $\times10^6\,\mathrm{yr}$ ) — and cloud counts match these lifetimes. But for an individual object the mapping is imperfect, because the measured SED depends on viewing geometry as much as age: an edge-on Class II disk can masquerade as embedded, and an object seen down its outflow cavity can look more evolved. So the sequence is a reliable clock for populations and a good first guess for individuals, but confirming a single source needs extra tracers — envelope mass, kinematics, outflow strength.

What sets the different lifetimes of Classes 0, I, and II?

Each phase is limited by different physics. Class 0 ( $\sim10^4\,\mathrm{yr}$ ) is set by the rapid free-fall infall of the densest gas, fast because $t_{\mathrm{ff}}\propto\rho^{-1/2}$ is short at high density. Class I ( $\sim10^5\,\mathrm{yr}$ ) is how long it takes to accrete and disperse the residual envelope, accelerated by widening outflow cavities. Class II (a few $\times10^6\,\mathrm{yr}$ ) is the disk-dissipation time, set by viscous accretion, ultraviolet and X-ray photoevaporation, and the locking of solids into planets. The declining infrared-excess fraction as clusters age independently confirms this few-Myr disk lifetime.

Does magnetic field change this picture of collapse?

Yes, substantially, though the classification survives. Magnetized cores resist collapse and, worse, threaten to carry off so much angular momentum that no disk forms — the magnetic braking catastrophe of ideal-MHD collapse. Reality is saved by non-ideal effects: ambipolar diffusion, Ohmic dissipation, and the Hall effect let neutral gas slip past the field lines in the dense interior, weakening the braking enough to permit the $\sim10$ – $100\,\mathrm{AU}$ disks actually observed. Fields also help launch and collimate the outflows that clear the envelope. So the SED-based sequence still holds, but the details of disk size, accretion, and outflow are set by magnetohydrodynamics, not gravity and pressure alone.