Galactic Halo and Globular Cluster System

The halo is now a precision laboratory, and Gaia is its instrument. By measuring parallaxes, proper motions, and — with spectroscopic surveys like APOGEE, H3, and DESI — full six-dimensional phase-space coordinates for millions of halo stars, astronomers have turned the Jeans and virial estimators from order-of-magnitude arguments into constraints on the Milky Way's mass good to tens of percent, converging near $M_{200}\sim 10^{12}\,M_\odot$. The same data broke the field open in 2018 with the Gaia-Sausage/Enceladus: a vast, radially-stretched cloud of stars in velocity space, the shredded remains of a dwarf galaxy of $\sim 10^{8}$–$10^{9}\,M_\odot$ that merged with the young Milky Way some 8–10 Gyr ago and supplied much of the inner halo and several globular clusters. Galactic archaeology now routinely separates in-situ from accreted populations using chemistry (the $[\alpha/\mathrm{Fe}]$ ratio) with orbits, reconstructing the Milky Way's merger tree. On the cluster side, the tidal streams of Palomar 5 and GD-1 have become sensitive probes of the dark halo's shape and lumpiness: gaps in a cold stream may record the gravitational wakes of dark subhalos, one of the few ways to test whether dark matter clumps on small scales as cold-dark-matter cosmology predicts. And globular-cluster ages, tightened by Gaia parallaxes, still stand as an independent lower bound on the age of the Universe, cross-checking the value from the cosmic microwave background.

Picture the halo as a swarm of bees around a hive. Each star follows its own long, plunging orbit with an essentially random orientation, so the cloud as a whole is nearly spherical and barely rotates. What holds it up against collapse is not ordered spin, as in the disk, but the sheer spread of stellar velocities — a pressure of random motion. Because those orbits are mostly radial, dropping steeply inward and climbing back out, the velocity ellipsoid is stretched along the line to the Galactic Centre; this radial anisotropy is a fingerprint of a halo grown by infall rather than settled from a spinning cloud. The first figure sets the stage: the luminous disk and bulge are small, dense features buried inside a far larger stellar halo of globular clusters, which is itself buried inside a dark-matter halo an order of magnitude larger again.

The globular clusters scattered through this cloud are ancient timestamps. Because the stars in a cluster all condensed together from one parcel of gas, a single isochrone fits the whole colour–magnitude diagram, and the luminosity of the main-sequence turnoff acts as a clock reading out the cluster's age. Their metal poverty tells us they formed when the Galaxy had barely begun to enrich itself with heavy elements. The decisive modern use of these tracers is quantitative: wherever we can measure their speeds we can weigh the mass interior to their orbits. That those speeds stay high at 30, 50, 100 kpc, where there is almost no light left to grip them, is a direct fingerprint of an unseen, massive dark halo. The second figure is the payoff — the enclosed mass climbing far past what the stars alone can account for.

Finally, the streams and "sausage" features that Gaia finds in velocity space are the still-cooling debris of swallowed dwarf galaxies — like cream stirred into coffee that has not yet fully mixed. Their survival as coherent structures tells us the halo was largely built by accretion and is dynamically young in mixing time, even though the individual stars are as old as the Universe.

The smooth stellar halo is described first by how its density thins outward. Star counts of RR Lyrae, blue horizontal-branch stars, and K giants show the number density falls as a power law in Galactocentric radius, steepening beyond a break near 25 kpc.

where: $\nu(r)$ — number density of halo stars at Galactocentric radius $r$; $n$ — power-law slope, observed near $2.5$–$3$ inside $\sim 25\,\mathrm{kpc}$ and steepening to $3.5$–$4.5$ beyond a break. Reading: the halo has no sharp edge — it fades continuously, and the break radius marks where accreted debris begins to dominate over stars formed in situ. The dynamical workhorse is the spherical Jeans equation, the first velocity moment of the collisionless Boltzmann equation for a steady, spherical tracer population. It links what we can measure — the tracer density and its radial velocity dispersion — to what we want, the enclosed mass, with the anisotropy as the crucial nuisance term.

where: $\sigma_r$ — radial velocity dispersion of the tracer (halo stars or clusters); $\beta$ — velocity anisotropy parameter; $M(<r)$ — total gravitating mass (stars, gas, dark matter) inside $r$; $G$ — Newton's constant. Reading: the left side is the tracer's own "pressure gradient" plus its anisotropy; the right side is gravity — measure the former and you have the latter. The anisotropy $\beta$ quantifies how radial the orbits are, and it is the term that makes weighing the halo genuinely hard. It is defined from the ratio of tangential to radial dispersions.

where: $\sigma_r,\sigma_\theta,\sigma_\phi$ — velocity dispersions in the spherical polar directions; $\beta=0$ — isotropic orbits; $\beta\to 1$ — purely radial orbits; $\beta<0$ — tangentially biased. Reading: the Milky Way halo has $\beta\approx 0.5$–$0.9$ over much of its extent — a strongly radial velocity ellipsoid, the kinematic scar of its accretion origin. Solving the Jeans equation for a power-law tracer collapses all of this into a compact tracer mass estimator, the single most useful formula in the article.

where: $\gamma \equiv -\mathrm{d}\ln\nu/\mathrm{d}\ln r$ — density slope; $\alpha \equiv -\mathrm{d}\ln\sigma_r^2/\mathrm{d}\ln r$ — dispersion slope; $\beta$ — anisotropy, entering with a minus sign. Reading: for a fixed measured $\sigma_r$, a more radial orbit population (larger $\beta$) implies a smaller enclosed mass — the mass–anisotropy degeneracy that Gaia proper motions exist to break. For a self-gravitating globular cluster in equilibrium, the scalar virial theorem does the same job with even less machinery, tying the cluster's mass to its size and internal velocities.

where: $K$ — total kinetic energy; $W\simeq -\alpha_{\rm s} GM^2/R$ — gravitational energy with a structure constant $\alpha_{\rm s}$ of order unity; $\langle v^2\rangle$ — mean-square stellar velocity; $R$ — a characteristic (half-mass or gravitational) radius. Reading: a cluster's velocity dispersion and half-light radius alone give its dynamical mass, $10^{4}$–$10^{6}\,M_\odot$. A cluster's internal evolution is set by how fast two-body encounters redistribute energy, measured by the half-mass relaxation time relative to the crossing time.

where: $t_{\rm rh}$ — two-body relaxation time inside the half-mass radius; $N$ — number of stars; $t_{\rm cross}\sim R/\langle v^2\rangle^{1/2}$ — crossing time; $\ln(0.4N)$ — Coulomb logarithm. Reading: for a $10^{6}$-star cluster $t_{\rm rh}\sim 4\,\mathrm{Gyr}$, shorter than the age of the Galaxy — clusters are collisional inside even though the halo around them is not. The globular cluster luminosity function has a near-universal turnover that makes the cluster population itself a standard candle.

where: $M_V$ — peak (turnover) absolute $V$ magnitude of the roughly Gaussian luminosity function; $\sigma_{M_V}$ — its width. Reading: the turnover sits near the same absolute magnitude in galaxy after galaxy, so measuring the apparent turnover magnitude of another galaxy's clusters yields its distance. Finally, the Galaxy's tide sets the outer boundary of each cluster through the Jacobi (tidal) radius, beyond which stars are stripped away.

where: $r_t$ — tidal radius; $r_p$ — perigalactic distance; $M_c$ — cluster mass; $M_g(<r_p)$ — Galactic mass inside $r_p$; $e$ — orbital eccentricity. Reading: clusters on plunging, eccentric orbits are pruned hardest at perigalacticon, and the stripped stars stream away into the halo as tidal tails.

1 · Dynamical mass of a globular cluster (fully worked)

The globular cluster 47 Tucanae has a line-of-sight velocity dispersion $\sigma_{\rm los}\approx 11\,\mathrm{km\,s^{-1}}$ and a half-light radius $r_h\approx 4\,\mathrm{pc}$. Estimate its dynamical mass from the virial theorem.

Step 1. Write the scalar virial theorem $2K+W=0$ with $K=\tfrac12 M\langle v^2\rangle$ and $W\simeq -\tfrac35 GM^2/R$ for a roughly uniform sphere. Solving, $M \simeq \langle v^2\rangle R/(\tfrac35 G)$.

Step 2. Convert the observed one-dimensional dispersion to the mean-square speed. For an isotropic system $\langle v^2\rangle = 3\sigma_{\rm los}^2$, so $M \simeq 3\sigma_{\rm los}^2 R/(\tfrac35 G) = 5\,\sigma_{\rm los}^2 R/G$, taking $R\approx r_h$.

Step 3. Insert numbers with $G=4.30\times10^{-3}\,\mathrm{pc\,km^2\,s^{-2}\,M_\odot^{-1}}$: $M \simeq 5(11)^2(4)/(4.30\times10^{-3}) = 2420/4.30\times10^{-3}$. This gives $M\approx 5.6\times10^{5}\,M_\odot$, an order-of-magnitude match to detailed dynamical models of 47 Tuc (which give $\sim 7\times10^{5}\,M_\odot$; the difference is the order-unity structure constant hidden in $\tfrac35$ and in the choice $R\approx r_h$).

Step 4. Sanity check the collisional timescale: with $N\sim 10^{6}$ and $t_{\rm cross}\sim r_h/\sigma_{\rm los}\sim 4\times10^{5}\,\mathrm{yr}$, equation (6) gives $t_{\rm rh}\approx 0.14N/\ln(0.4N)\times t_{\rm cross}\approx 4\,\mathrm{Gyr}$. Because $t_{\rm rh}<t_{\rm Hubble}$, 47 Tuc is dynamically relaxed — mass-segregated, with a dense evolved core.

2 · Weighing the Milky Way inside 30 kpc (fully worked)

Halo stars at $r=30\,\mathrm{kpc}$ show a radial velocity dispersion $\sigma_r\approx 120\,\mathrm{km\,s^{-1}}$, a density slope $\gamma\approx 3.5$, a mildly falling dispersion $\alpha\approx 0.5$, and anisotropy $\beta\approx 0.5$. Find the enclosed mass and the dark-matter fraction.

Step 1. Form the coefficient $\gamma+\alpha-2\beta = 3.5+0.5-2(0.5) = 3.0$.

Step 2. Apply the estimator (4): $M(<r) = \sigma_r^2 r(\gamma+\alpha-2\beta)/G$. With $G=4.30\times10^{-6}\,\mathrm{kpc\,km^2\,s^{-2}\,M_\odot^{-1}}$, $M = (120)^2(30)(3.0)/(4.30\times10^{-6})$.

Step 3. Evaluate: $(120)^2=1.44\times10^{4}$; times $30$ is $4.32\times10^{5}$; times $3.0$ is $1.30\times10^{6}$; divided by $4.30\times10^{-6}$ gives $M(<30\,\mathrm{kpc})\approx 3.0\times10^{11}\,M_\odot$.

Step 4. Compare to the stellar mass. The Galaxy contains only $\sim 6\times10^{10}\,M_\odot$ in stars, most of it already enclosed at 30 kpc. The dark fraction is $(3.0\times10^{11}-6\times10^{10})/(3.0\times10^{11})$, so about 80% of the mass inside 30 kpc is dark. This is the halo tracers doing exactly what they are for.

3 · Tidal radius and stream formation

A cluster of mass $M_c=6\times10^{5}\,M_\odot$ reaches perigalacticon at $r_p=10\,\mathrm{kpc}$, where the Galaxy encloses $M_g\approx 1\times10^{11}\,M_\odot$. Find its tidal radius.

Step 1. Use the circular-orbit Jacobi radius $r_t \simeq r_p(M_c/3M_g)^{1/3}$.

Step 2. Evaluate the ratio: $M_c/3M_g = 6\times10^{5}/(3\times10^{11}) = 2\times10^{-6}$, whose cube root is $1.26\times10^{-2}$.

Step 3. So $r_t \simeq 10\,\mathrm{kpc}\times1.26\times10^{-2} = 0.126\,\mathrm{kpc}$, or $r_t\approx 126\,\mathrm{pc}$ — dozens of times the cluster's half-light radius, yet stars beyond it are stripped at each perigalactic passage, unwinding into the leading and trailing tidal tails that appear as halo streams.

Before anyone could weigh the Galaxy with globular clusters, someone had to realize the clusters were not scattered at random — and that they hold the key to where we sit inside the Milky Way.

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 24–25 (the Milky Way, the halo, and globular clusters) · Binney and Tremaine, Galactic Dynamics, 2nd ed., Ch. 4 (equilibria and the Jeans equations) and Ch. 8 (collisionless systems and tides) · Mo, van den Bosch and White, Galaxy Formation and Evolution, Ch. 11 and 13 (dynamics and the stellar halo) · Ryden, Introduction to Cosmology, and Dodelson, Modern Cosmology (hierarchical structure formation).

Provenance. The halo density slopes, the $M_V\approx -7.5$ luminosity-function turnover, cluster ages of $11$–$13\,\mathrm{Gyr}$, and the metallicity bimodality are standard values from Carroll and Ostlie Ch. 24–25. The Jeans, virial, relaxation, and Jacobi-radius relations follow Binney and Tremaine Ch. 4 and 8. A verifier should recompute: (1) $M(<30\,\mathrm{kpc})\approx 3\times10^{11}\,M_\odot$ from $\sigma_r=120\,\mathrm{km\,s^{-1}}$, $r=30\,\mathrm{kpc}$, coefficient $3$, and $G=4.30\times10^{-6}\,\mathrm{kpc\,km^2\,s^{-2}\,M_\odot^{-1}}$; (2) the 47 Tuc virial mass $\sim 6\times10^{5}\,M_\odot$; (3) the tidal radius $r_t\approx 126\,\mathrm{pc}$; (4) the mass–lifetime scaling $t_{\rm MS}\propto M^{-2.5}$. The accretion-history context follows Searle–Zinn and the modern Gaia literature.

Prerequisites. Structure of the Milky Way Galaxy · Stellar Populations I, II, and III · Galactic Coordinate System and the Solar Neighborhood · Galactic Rotation Curves. Leads to. Evidence for Dark Matter in Galaxies · Galactic Dark Matter Halos and Density Profiles · Elliptical Galaxies and the de Vaucouleurs Profile.

Ask the tutor: Want to see how Gaia proper motions break the mass–anisotropy degeneracy, or how a tidal stream like GD-1 turns into a probe of dark-matter subhalos?