Thick & Thin Disk

Two disks in one plane

Edge-on photographs of spiral galaxies almost always show a thin luminous wafer split by a dark dust lane — and, fainter, a slightly fatter glow above and below it. That second glow is not an artefact. Our own Milky Way is built the same way: the disk you would draw as a single flat pancake is actually two interleaved populations of stars occupying the same plane but with very different thicknesses, ages, chemistry, and orbits. We call them the thin disk and the thick disk.

The thin disk is where almost everything you can name lives — the Sun, the spiral arms, the open clusters, the molecular clouds, the ongoing star formation. It is geometrically thin (a vertical scale height of order 300 pc) and dynamically cold: its stars stay close to the midplane on near-circular orbits. The thick disk is a more diffuse, more spheroidal layer that reaches roughly three times farther from the plane. Locally it is rare — only a few percent of stars in the solar neighborhood belong to it — but its stars carry a chemical fingerprint marking them as among the first the Galaxy ever made. The two disks overlap completely in space: a star found 200 pc above the plane could belong to either, and you cannot tell from its position alone.

The vertical density profiles

The cleanest way to see two disks is to count stars as a function of height z above the Galactic plane. The combined number density is well fit by a sum of two components — a self-gravitating isothermal sheet for the thin disk and an exponential for the thick disk:

ρ(z) = ρ_thin · sech²(z / 2h_thin)  +  ρ_thick · exp(−|z| / h_thick)

h_thin  ≈ 300 pc        (thin-disk scale height)
h_thick ≈ 900 pc        (thick-disk scale height)
ρ_thick / ρ_thin |_(z=0)  ≈ 2–10 %   (local normalization)

The sech² form is the exact solution for an isothermal, self-gravitating slab; far from the plane it tends to a simple exponential, so both terms decline roughly exponentially at large z. Because the thick disk falls off three times more slowly, it eventually wins: above about 1.5 kpc the star counts are dominated by thick-disk members even though, at the midplane, the thin disk outnumbers them ten to one or more. The crossover near 1 kpc is exactly the "kink" in the star-count slope that revealed the second component.

How the split was discovered

In 1983 Gerry Gilmore and Neill Reid published deep photographic star counts toward the South Galactic Pole — a direction that looks straight out of the disk, so height above the plane maps directly to distance. A single exponential layer cannot fit those counts: the observed density falls off steeply close to the plane (the thin disk) but then flattens into a shallower decline at greater heights. They modeled the excess as a second exponential with a scale height of about 1.3 kpc (modern estimates pull it down toward 0.9 kpc) and a local density a few percent that of the thin disk. That second term is the thick disk. The interpretation was not universally accepted at first — some argued it was just the high-z tail of the thin disk — but follow-up work on the stars' chemistry and kinematics confirmed they are a genuinely distinct population, not a smooth extension.

Chemistry: the α-bimodality

The decisive evidence is chemical. Alpha elements — oxygen, magnesium, silicon, calcium, titanium — are forged almost entirely by core-collapse (Type II) supernovae, which detonate within a few million years of a massive star's birth. Iron, by contrast, is added more gradually: roughly half of the Galaxy's iron comes from Type Ia supernovae, white-dwarf explosions that take on the order of 1 Gyr to occur. So the ratio [α/Fe] is a clock. A population that forms quickly, before the Type Ia iron arrives, locks in a high [α/Fe]; a population that forms slowly, sipping iron the whole time, ends up near solar [α/Fe] ≈ 0.

When large spectroscopic surveys (RAVE, then SDSS/APOGEE, then GALAH) measured [α/Fe] versus [Fe/H] for hundreds of thousands of stars, they found not a smooth cloud but two distinct sequences separated by a real gap:

• The high-α sequence — α-enhanced ([α/Fe] ≈ +0.2 to +0.4), spanning low to moderate metallicity ([Fe/H] from about −1.0 up to −0.1). This is the thick disk. The enhancement says it formed fast and early.
• The low-α sequence — near-solar [α/Fe], spanning [Fe/H] from about −0.5 to +0.5. This is the thin disk, built slowly over many gigayears as Type Ia iron steadily diluted the α-elements.

Crucially, the gap means a star's chemistry classifies it far more cleanly than its position or velocity. Two stars at the same height with the same metallicity can sit on opposite sides of the α gap — one thin, one thick. This chemical definition has largely superseded the older geometric one.

The numbers side by side

Property	Thin disk	Thick disk
Vertical scale height	≈ 300 pc	≈ 900–1000 pc
Radial scale length	≈ 2.6 kpc	≈ 2.0 kpc (more centrally concentrated)
Typical stellar age	0 – 8 Gyr	10 – 12 Gyr
Median metallicity [Fe/H]	≈ −0.1	≈ −0.6 (range −1.0 to −0.2)
α-enhancement [α/Fe]	≈ 0 (near solar)	≈ +0.3 (enhanced)
Vertical velocity dispersion σ_z	≈ 20 km/s	≈ 40–50 km/s
Rotation relative to LSR	≈ in step (lag ≲ 5 km/s)	lags by ≈ 50 km/s
Local density fraction (z = 0)	~95–98 %	~2–5 %
Stellar mass (Milky Way)	~4 × 10¹⁰ M☉	~few × 10⁹ M☉

The pattern is internally consistent: the thick disk is older, poorer in metals, richer in α-elements, hotter kinematically (it puffs up because its stars have larger vertical excursions), and rotates more slowly. Every one of these properties points the same way — toward an early, rapid formation followed by dynamical heating or by birth in a hotter state.

Kinematics: why thick means hot

"Thick" is ultimately a statement about orbits. The vertical extent of a population is set by how much vertical energy its stars carry. To first order, the scale height and the vertical velocity dispersion are linked through the local vertical gravity: a star oscillates through the plane with an amplitude that grows with its vertical velocity. Roughly,

σ_z² ≈ 2π G Σ h          (isothermal sheet, surface density Σ, scale height h)
⇒  h ∝ σ_z²              (for a fixed background surface density Σ)

Thin:  σ_z ≈ 20 km/s  →  h ≈ 300 pc
Thick: σ_z ≈ 35 km/s  →  h ≈ 900 pc   (σ_z larger by ~√3 ≈ 1.7× → h larger by ~3×)

The relationship between dispersion and thickness is roughly quadratic at fixed surface density (h ∝ σ_z²), so it only takes a factor ~1.7 rise in vertical velocity to triple the scale height. In practice the thick disk's measured vertical dispersion is even larger, σ_z ≈ 40–50 km/s, because its stars range to heights where the restoring gravity weakens and the simple fixed-Σ scaling underestimates the speeds needed — but the message is the same: a modestly hotter population puffs up dramatically. The thick disk's stars also lag the local standard of rest by about 50 km/s. This asymmetric drift follows from the same large random motions: a hotter population needs less ordered rotation to balance against gravity, so on average it orbits more slowly. The thin disk, dynamically cold, is nearly in pure circular rotation.

Where the thick disk came from

Why does the Galaxy have a separate hot, old layer at all? Four families of explanation compete, and they are not mutually exclusive:

• Heating of a pre-existing thin disk. A merging satellite galaxy, or repeated scattering off spiral arms and giant molecular clouds, pumps random energy into disk stars over time, puffing an originally thin disk into a thick one. Secular GMC/spiral heating is too gentle to make the full thick disk on its own, but a single significant merger can do it in one event.
• Born thick (gas-rich turbulent disk). At high redshift, disks were gas-rich and clumpy, with enormous turbulent velocities. Stars forming in such a disk are born hot — the thick disk could simply be the fossil of the Galaxy's first, violent star-forming epoch, with no heating required afterward.
• Accretion of satellite stars. Some thick-disk stars may be debris from disrupted dwarf galaxies whose orbits were dragged into the plane. Pure accretion struggles to reproduce the thick disk's tight chemical sequence, so this is at most a contributor.
• Radial migration. Resonant scattering off spiral arms can move stars to new galactocentric radii without heating them much; old inner-disk stars migrating outward can mimic a thicker, older population in the solar neighborhood.

The modern consensus leans on a key clue: the thick disk's α-enhanced sequence terminates at a metallicity and age that line up with the last major merger the Milky Way experienced, the Gaia-Sausage-Enceladus event about 10 Gyr ago, identified from Gaia astrometry in 2018. The favored story is that the early Galaxy formed a gas-rich, turbulent disk that was thickened around the time of that merger; afterward, the gas settled and the thin disk grew quietly on top, slowly enriching toward solar α/Fe.

What it looks like in real galaxies

• The Milky Way. Gaia parallaxes plus APOGEE and GALAH spectroscopy have mapped both disks star by star. The α-[Fe/H] bimodality, the age split (thin: ≲8 Gyr; thick: 10–12 Gyr), and the kinematic lag are all now measured directly in the solar neighborhood and out to several kiloparsecs.
• Edge-on spirals. Deep imaging of galaxies like NGC 891 and NGC 4565 shows a clearly resolved thick-disk component — a faint, vertically extended, redder light distribution surrounding the thin star-forming midplane. Star-count studies of the nearby Andromeda Galaxy (M31) likewise find a chemically and kinematically distinct thick disk.
• The α-gap as a galactic clock. Because the thick disk records the rapid early phase and the thin disk the slow later phase, the two-sequence chemistry is now a standard tool for reconstructing a galaxy's star-formation history — galactic archaeology in its most literal form.
• The Sun's place. The Sun sits about 20 pc above the midplane and is a thin-disk star, ~4.6 Gyr old with [Fe/H] ≈ 0 and near-solar α/Fe. It is firmly in the low-α sequence — a member of the younger, quieter disk.

Common misconceptions and edge cases

• "The thick disk is just the high-z tail of the thin disk." This was the original objection to Gilmore & Reid, and it is wrong: the two populations differ in age, chemistry, and kinematics, not merely in height. A smooth single disk cannot reproduce the α-[Fe/H] gap.
• Confusing the thick disk with the stellar halo. The halo is a roughly spherical, far more metal-poor ([Fe/H] ≲ −1.5), pressure-supported component with little net rotation. The thick disk still rotates (it merely lags by ~50 km/s) and is more metal-rich and flattened. They overlap in metallicity but separate cleanly in kinematics and spatial shape.
• Assuming you can classify a star by its height. Because the disks overlap completely in space, geometric or even kinematic selection is statistical and contaminated. Chemistry — the α gap — is the reliable discriminant.
• Treating "thick disk" as one age. The high-α sequence spans a range of metallicities and a couple of gigayears of age; it is a rapidly enriched population, not a single coeval burst. Its defining feature is the elevated α/Fe, not a single birthday.
• Forgetting the metal-weak thick disk. A tail of α-enhanced stars extends down to [Fe/H] ≈ −1.0 and below — the "metal-weak thick disk" — which can be mistaken for halo stars unless their rotation and α-enhancement are measured.
• Reading the local density fraction as the mass fraction. The thick disk is only a few percent of stars at the midplane, but because it extends much farther from the plane and is centrally concentrated, its total stellar mass is closer to ~10–20% of the disk — far more than the midplane count suggests.