Vertical structure and dynamics of a galactic disk

Chanda J. Joga,
a
Department of Physics, Indian Institute of Science, Bangalore 560012, India.
arXiv:2507.02062v1 [astro-ph.GA] 2 Jul 2025

Abstract
Most of the visible mass in a typical spiral galaxy is distributed in a thin disk,
with a radial extent much larger than its thickness. While the planar disk
structure, including non-axisymmetric features such as spiral structure, has
been studied extensively, the vertical structure has not received comparable
attention. This review aims to give a comprehensive, pedagogic introduction
to the rich topic of vertical structure of a galactic disk in hydrostatic equi-
librium, and discuss the theoretical developments in this field in the context
of recent observations. A realistic multi-component disk plus halo model of
a galaxy has been developed and studied by us in detail. This takes account
of both stars and interstellar gas, treated as isothermal components with
different velocity dispersions, which are gravitationally coupled; further, the
disk is in the gravitational field of the dark matter halo. This review focuses
on this model and the results from it in different physical cases.
The gas and halo crucially affect the resulting self-consistent stellar dis-
tribution such that it is vertically constrained to be closer to the mid-plane,
and has a steeper profile than in the standard one-component case, in agree-
ment with modern observations. A typical stellar disk is shown to flare by a
factor of few within the visible radial extent of the disk. These robust results
question the sech2 profile and a constant scale height, routinely used in the
literature for convenience. In an important application, the observed HI gas
scale height is used as a constraint on the model which helps determine the
shape and the density profile of the dark matter halo for galaxies. Finally, we
outline some key, open questions which can be addressed in the near future
using the above model, and new observational data – for example, from the
IFU surveys and JWST. These promise to give a better understanding of
this topic.

Email address: cjjog@iisc.ac.in (Chanda J. Jog)

Preprint submitted to Elsevier July 4, 2025

Keywords: Galaxies: structure, Galaxies: kinematics and dynamics,
Galaxies :ISM, Galaxies: halos, Galaxy: disk, Galaxy: structure

Contents

1 Introduction 4
1.1 Background and motivation . . . . . . . . . . . . . . . . . . . 4
1.2 Outline of the review . . . . . . . . . . . . . . . . . . . . . . . 7

2 Observations 10
2.1 Vertical stellar density distribution . . . . . . . . . . . . . . . 10
2.1.1 Analysis of intensity profiles . . . . . . . . . . . . . . . 11
2.1.2 Analysis of 3-D light distribution . . . . . . . . . . . . 15
2.1.3 Analysis of star counts data in the Galaxy . . . . . . . 16
2.1.4 Stellar velocity dispersion data . . . . . . . . . . . . . 17
2.2 Interstellar gas: Surface density and velocity dispersion . . . . 18

3 Self-consistent vertical density distribution in a galactic disk 19

3.1 Basic, classical model of a galactic disk . . . . . . . . . . . . . 20
3.1.1 A brief comment about the Oort limit . . . . . . . . . 25
3.2 Gas in a galactic disk in the classical treatment . . . . . . . . 27
3.3 Vertical constraining effect of molecular cloud complex on disk 28

4 Multi-component disk plus halo model 33

4.1 Motivation and basic outline of the model . . . . . . . . . . . 34
4.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.1.2 Basic outline of the model . . . . . . . . . . . . . . . . 36
4.2 Multi-component disk plus halo model: Thin disk . . . . . . . 38
4.2.1 Formulation of equations: Thin disk . . . . . . . . . . . 38
4.2.2 Numerical solutions . . . . . . . . . . . . . . . . . . . . 41
4.2.3 Application to the Milky Way . . . . . . . . . . . . . . 42
4.2.4 Results for vertical scale heights in the inner Galaxy . 44
4.2.5 Physical understanding of re-distribution of stellar den-
sity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2.6 Detailed results for stellar density distribution in the
outer Galaxy . . . . . . . . . . . . . . . . . . . . . . . 54
4.2.7 Application to NGC 891, NGC 4565: Radial variation
in stellar scale height . . . . . . . . . . . . . . . . . . . 62

2
 4.3 Multi-component disk plus halo model: Thick, or low density
disk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.3.1 Inclusion of radial term in the disk Poisson equation . . 67
4.3.2 Thick, or low density disk case: Applications . . . . . . 73
4.4 General models . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.4.1 Model for thick, or low density disk, using complete
Jeans equations . . . . . . . . . . . . . . . . . . . . . . 83
4.4.2 Model for a non-isothermal galactic disk . . . . . . . . 88
4.5 Applications to observational data and theoretical studies in
the literature . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.5.1 Applications to analyze intensity profiles . . . . . . . . 97
4.5.2 Applications to the study of gas distribution . . . . . . 98
4.5.3 Applications to theoretical studies . . . . . . . . . . . . 99
4.5.4 Comments on simplified, non-self-consistent prescrip-
tions for stellar and HI scale heights in the literature . 100
4.6 Gravitational potential energy of a multi-component galactic
disk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.6.1 Motivation and Formulation . . . . . . . . . . . . . . . 104
4.6.2 Potential energy for a disk in hydrostatic equilibrium . 105
4.6.3 Gravitational potential of a coupled disk . . . . . . . . 108

5 Flaring stellar disk in a galaxy: A generic result 110

5.1 Physical basis for the radial variation in vertical thickness . . 111
5.1.1 Dependence on radial variation of stellar velocity dis-
persion . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.2 Origin of sharp flaring in the outer disk . . . . . . . . . . . . . 113
5.3 Application to galaxies: Theoretical modeling and observa-
tional results . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.3.1 The Milky Way . . . . . . . . . . . . . . . . . . . . . 115
5.3.2 UGC 7321 . . . . . . . . . . . . . . . . . . . . . . . . 117
5.3.3 NGC 891, NGC 4565 . . . . . . . . . . . . . . . . . . 118
5.3.4 Other galaxies . . . . . . . . . . . . . . . . . . . . . . 118
5.3.5 Evidence of flaring from numerical simulations . . . . 119
5.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.5 Dynamical implications, and future directions . . . . . . . . . 121

6 Density profile and shape of dark matter halo 123

6.1 Background on dark matter halos in galaxies . . . . . . . . . . 123

3
 6.2 Motivation for using HI gas scale height as a constraint . . . . 124
6.3 Dark matter halo traced using HI gas scale height: Self-consistent
approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.3.1 Application to the Milky Way . . . . . . . . . . . . . . 128
6.3.2 Application to M31 (Andromeda) . . . . . . . . . . . . 133
6.3.3 Application to UGC 7321 . . . . . . . . . . . . . . . . 134
6.4 Dark matter halo traced using HI scale heights: Non-self-
consistent models in the literature . . . . . . . . . . . . . . . . 136
6.5 Discussion and Future directions . . . . . . . . . . . . . . . . . 138
6.6 Shape of halo and its effect on disk dynamics . . . . . . . . . . 140
6.7 Other tracers and complementarity . . . . . . . . . . . . . . . 141
6.8 MOND (MOdified Newtonian Dynamics) as an alternative to
Dark matter halo . . . . . . . . . . . . . . . . . . . . . . . . . 142

7 Related topics 144

7.1 Thick disk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
7.2 Dynamical features along the vertical direction . . . . . . . . . 146
7.2.1 Bending instabilities leading to warps, and corrugations 147
7.2.2 Breathing motions normal to the galactic disk . . . . . 150
7.2.3 Gaia Phase-space spiral . . . . . . . . . . . . . . . . . 152

8 Vertical structure and multi-component disk plus halo model:

Future directions 155

9 Conclusions 160
9.1 Main results from the multi-component disk plus halo model . 161
9.2 Dynamical implications of the model . . . . . . . . . . . . . . 164
9.3 Unique features of study of disk vertical structure . . . . . . . 165
9.4 Future problems and trends . . . . . . . . . . . . . . . . . . . 166
9.5 Summary and outlook . . . . . . . . . . . . . . . . . . . . . . 167

10 Appendix A. Thickness of a self-gravitating disk and disk

vertical density profile 171

1. Introduction
1.1. Background and motivation
A glance at the image of a spiral galaxy inclined close to edge-on w.r.t.
the line of sight (e.g. NGC 891) tells us that most of the visible mass of

4
the galaxy is distributed in a thin disk, with the radial extent much larger
than the vertical thickness [1]. For example, the ratio of the thickness to the
radius of the Milky Way is ∼ 1/40 [2]. Most of the mass in a galactic disk is
in stars, and a small fraction (∼ 10 − 15%) is contained in interstellar gas.
The disk distribution is the most striking feature of a galaxy: see Fig. 1 for
images of four typical examples of galaxies that are viewed edge-on. Further,
the rotation in a galaxy can be measured which is not strictly face-on, this
indicates an underlying disk distribution [3, 4]. The rotation in the Milky
Way was first clearly measured using the 21 cm line emission from HI gas
[5]. For these two reasons, spiral galaxies are also sometimes referred to as
disk galaxies.
The vertical structure of a galactic disk is a tracer of disk potential; hence,
it is crucial to know the vertical structure in order to understand the disk
dynamics and evolution. Edge-on galaxies play a crucial role in the study
of disk vertical structure. The thin disk distribution of matter in a galaxy
simplifies its dynamics, because the dynamics in the vertical direction can
then be taken to be de-coupled from that in the plane ([2]; also, see Section
3.1).1 The disk is supported against gravitational collapse in the plane mainly
by rotation w.r.t. the disk centre; whereas it is supported against collapse in
the vertical direction by its internal pressure gradient. While the planar disk
structure, including non-axisymmetric features such as spiral structure, has
been studied extensively; the vertical structure has not received comparable
attention. This is partly due to the observational challenges of studying the
disk vertical structure (Section 2); and also due to the necessity of having to
theoretically solve for the entire self-consistent vertical distribution at a go
(Sections 3, 4).
In recent years, an explosion of astrometric data with unprecedented de-
tail for the position and kinematics of over a billion individual stars in the
Milky Way has become available: from Gaia, LAMOST, APOGEE and other
telescopes. This gives very detailed information on the disk vertical struc-
ture, and allows us to determine the vertical distribution in situ. This has to

1
However, the physical origin of the thinness of the disk, and the origin of the disk itself,
is not yet fully understood; although the conservation of angular momentum is believed
to be the reason for this [6, 7]. During the formation of a galaxy, the gas clouds cannot
collapse easily in a direction normal to the direction of the angular momentum vector –
due to the conservation of the angular momentum – but they can collapse readily along
it. Hence the collapse leads to a disk.

5
Figure 1: A block of four typical edge-on galaxies: NGC 891 (top left panel); NGC 4565
(top right panel); NGC 4013 (lower left panel); and NGC 5907 (lower right panel). An
edge-on galaxy is by chance so oriented that the l.o.s. from the observer cuts across the
disk plane. This brings out the thin nature of the disk. The dark band cutting across the
middle of each image is due to absorption by dust associated with the interstellar gas in
the galactic disk. Edge-on galaxies provide crucial information for the study of vertical
structure of galactic disks. (Courtesy: NOAO)

6
be understood and interpreted theoretically for an optimal use of the data.
Although isolated papers in the literature address some issues related to
the disk vertical structure, there is so far no cohesive review of this fast-
growing, active field. Given these reasons, a review article focused on the
vertical structure of a galactic disk is warranted at this juncture.
Some noteworthy previous reviews on related topics are: about galaxy
disks in general [8] ; and the structure and kinematics of the Milky Way [9].
Two recent reviews deal with the dynamics and evolution of the Milky way,
particularly in light of the Gaia observations [10, 11]. However, these do not
specifically focus on the vertical structure of a galactic disk in detail.

1.2. Outline of the review

In this review, we aim to give a comprehensive introduction to the rich
topic of vertical structure and related dynamics of a typical, thin galactic
disk, in the context of recent observations. We will present this subject
starting from the early observations and theoretical work on this in the lit-
erature, followed by more recent developments, in a pedagogical fashion; to
bring out the physics involved clearly.
The study of the vertical structure of a galactic disk has a long his-
tory. The classic work by Spitzer [12] considered an idealized self-gravitating,
isothermal, locally dense disk2 which was shown to obey a sech2 (or a hyper-
bolic secant squared) form (also, see [13]; and Section 3).
However, with the advent of R-band studies of external galaxies in early
1980’s, which allowed studies of underlying old stellar mass distribution; it
was already clear that the distribution close to the mid-plane showed an
excess over that expected for a sech2 distribution [14, 15]. These profiles
could instead be better fit by a steeper functional form, such as a sech or
an exponential. A somewhat ad hoc model was proposed by [14] to fit this
observed trend. Surprisingly, no further work was done on this problem for
a long time.
In recent years, the problem of disk vertical structure has been addressed
in a more physically motivated approach, starting from the work by Narayan
& Jog (2002) [16], and later followed by other studies [17, 18, 19] . The
multi-component disk plus halo model first proposed and worked out by [16],

2
We discuss in Section 3.1 how this is equivalent to treating a thin disk – for which the
result by [12] is routinely applied.

7
treats a realistic galactic disk as a multi-component system, where the stars
and interstellar gas (HI and H2 ) are taken to be gravitationally coupled; and
the disk is taken to be in the gravitational field of the dark matter halo.
Thus, this model constitutes a more complete physical treatment of the disk
vertical structure than the stars-alone case that is usually considered; and it
also naturally explains various observational features of the vertical density
distribution of stars and gas. In this review, we focus on this model and the
results from it.
In the above model [16], the self-consistent vertical density distribution
of each disk component is obtained simultaneously by solving the coupled,
joint Poisson-hydrostatic balance equations for all the three disk components.
The resulting vertical density distribution is obtained in terms of the physical
parameters such as the surface density and velocity dispersion of each disk
component, and the halo mass distribution.
The most significant finding is that the gravitational force due to the in-
terstellar gas and the dark matter halo significantly affects the self-consistent
vertical stellar distribution in the disk. The modified stellar distribution is
constrained closer to the mid-plane, such that: it has a higher mid-plane
density, a smaller scale height, and is steeper than the corresponding stars-
alone case. The constraining effect due to the gas and the halo dominate in
the inner and the outer Galaxy, respectively. The resulting smaller vertical
thickness for all three disk components agree well with observations for the
Milky Way [16, 19]. The resulting stellar density distribution is steeper than
sech2 , as observed [18, 19].
The vertical structure and dynamics of a galactic disk is a complex and
challenging subject. We will discuss various physical cases that deal with
progressively more detailed physical points, and tie these in with real astro-
physical conditions. On inclusion of some of these realistic physical effects,
such as a non-isothermal velocity dispersion, the resulting vertical density
distribution can differ substantially, by ∼ 30 to 40 % , compared to the sech2
distribution obtained in the standard isothermal, one-component, thin disk
case [20, 21].
A generic, robust result obtained is that for a typical stellar disk in hy-
drostatic equilibrium, the resulting scale height increases by a factor of few
within the optical disk.3 The flaring stellar disk is in contrast to a constant

3
The radius of the optical disk or the visible extent of the stellar disk is typically defined

8
scale height commonly assumed in the literature [22]. However, such a trend
of a flaring disk is in agreement with recent observations and simulations.
Further, in an important application, the observed HI gas scale heights
and the rotation curve have been used as two simultaneous constraints on the
above model to determine the shape and the density profile of the dark matter
halo. This approach has been applied to a number of galaxies, including the
Milky Way and M31.
We mention that this review will focus on the ”canonical” thin disk. We
do not treat the chemically and kinematically distinct ”Thick disk” compo-
nent that is common in many galaxies. However, we will discuss the latter
briefly for the sake of completeness (see Section 7.1; also, see Sections 4.5.1,
and 8).
The study of disk vertical structure is an active field, with many new,
important questions – mainly triggered by input from new observations. For
example, critical input parameters such as the stellar velocity dispersion in
external galaxies can now be extracted from the near-IR, kinematical IFU
data. Using these, more accurate model vertical density distribution for
galaxies can be determined. The model can be also applied to understand
the vertical structure of high redshift galaxies; for example, using the data
from JWST (James Webb Space Telescope). Further, the overall vertically
constrained stellar density distribution in a coupled case can have important
dynamical consequences: it can make the disk more stable against distortion;
and also affect dynamical features such as warps. These effects need to be
studied.
Finally, apart from being interesting in itself, the vertical structure of
a galactic disk is important because it holds clues to the formation and
evolution of the disk [23, 24]. The evolution, including the increase in the
disk scale height, could be either due to internal or secular processes, such
as heating by spiral arms and bars; or, due to external sources such as tidal
encounters with external galaxies (see [2, 23, 24], and references therein).
A section-wise break-up of the contents of this review is as follows. In
Section 2 , we describe the observational challenges of studying the vertical
structure, and the data obtained. Section 3 mainly discusses the classic
work by Spitzer [12]. In Section 4, we present the modern multi-component

to be the radius corresponding to a limiting isophote: say, 26.5 mag arc sec−2 , which is
known as the Holmberg radius. See Appendix A for details.

9
disk plus halo model, and its applications, to cases with different physical
conditions: such as, a thin disk; a thick or a low density disk – which contain
progressively more detailed physics in the problem. These cases represent
different regions and types of galaxies. Sections 5 and 6 present and explain
a flaring stellar disk as a generic result; and how the shape and the density
profile of dark matter halo are deduced, respectively. These two sections
could be read as stand-alone mini-reviews on these two topics. Section 7 gives
a brief introduction to related topics such as thick disks, bending instabilities,
and phase-space spirals. Section 8 outlines some specific, well-defined, open
problems related to the disk vertical structure, and future trends in the field.
Section 9 summarizes the conclusions from this review. In Appendix A,
various physics issues related to the thickness of a self-gravitating disk and
the disk vertical density profile as presented in the review are put together
and discussed. At the end, a glossary of frequently used physical terms is
defined in Table 1; also, a list of frequently used acronyms is given.

2. Observations
The observations of the vertical intensity profiles for an edge-on galaxy;
the density distribution and velocity dispersion of stars and gas in a galactic
disk, will be given here briefly. These provide background about the topic;
also, these are necessary to help formulate theoretical models of the disk
vertical structure. This section aims to give an idea of the basic issues and
complexities involved in observations related to the disk vertical structure.

2.1. Vertical stellar density distribution

An image of any edge-on galaxy clearly shows that the light, and hence
the stellar mass distribution, in a galaxy is mainly in a thin disk. However,
to determine the vertical density distribution from the observed intensity
distribution in external galaxies is problematic because of various effects:
such as, the contamination due to projection effects from non-local regions in
the disk along the line of sight, and dust extinction, as discussed next. In our
Galaxy, an alternative approach of actual star counts is possible in the solar
neighbourhood. This allows an in situ measurement of density distribution
in great detail, as will be discussed later in this Section (Section 2.1.3). This
approach was first used in the classic work by Oort [25] to determine the
mid-plane density; the latter problem has now found resurgence with new
data from Gaia and other sources (see Sections 2.1.3, and 3.1.1).

10
2.1.1. Analysis of intensity profiles
The determination of vertical density distribution from observations is
a challenging task. This is why a basic theoretical model for the vertical
structure of a thin, self-gravitating disk was first developed ab initio [12, 13].
This was later used in the literature to analyze the observed intensity profiles,
so as to determine the model parameters (from [12]) for the vertical disk
density distribution, as discussed next. This is in contrast to most topics
in astronomy where the observations are used as a guideline to construct a
basic theoretical model of the system.
To study the vertical density profile in a galactic disk, one needs to look
at galaxies that are edge-on or tilted close to edge-on (or 900 ) w.r.t. line of
sight. However, due to projection effects, the intensity as observed from a
point R on the major axis (at which a line of sight cuts it) has contributions
from non-local regions along the line of sight. These regions are at different
radial distances from the galactic centre. Further, the light from a galactic
disk suffers an extinction due to interstellar dust. This effect is proportional
to the distance along the line of sight within the disk, since optical depth is a
cumulative property. The dust extinction is important at visible wavelengths
and becomes less at longer wavelengths. Therefore, the early detailed studies
of edge-on studies of galaxies were done using R-Band or I-band observations,
e.g.,[26], to minimize the effect of dust extinction. The modern studies use
near-IR data where the effect of dust extinction is further minimized, see for
example, the S4 G (Spitzer Survey of Spiral Galaxies) catalog [27].
Intensity profiles
Some of the pioneering work on this topic was done by van der Kruit $
Searle (1981) [22, 28]. Their aim was to study the vertical structure of a set
of edge-on galaxies by analyzing the three-dimensional distribution of light
in the galactic disk. From the observed photometric image of a galaxy, they
plotted intensity profiles along z at cuts taken at different radii along the
major axis ([22]). This procedure was later followed by many authors, and
applied for the study of other galaxies [15, 29].
To analyze the intensity profiles, [22] adopted a basic theoretical model
for the vertical density distribution as for an isothermal disk [12]; which
has a sech2 (z/z0 ) form, where z0 is an indicator of disk thickness (to be
discussed in Section 3.1). Further, they assumed a surface density that falls
off exponentially with radius (as observed, see [30]). Next, they calculated
the theoretical intensity profiles, I(R, z), where R is the distance from the

11
galaxy centre along the major axis, where the line of sight cuts the major
axis. For simplicity, all the stars were taken to form a single component.
These theoretical profiles were then compared with the observed intensity
profiles to deduce the model parameters (from [12]). This is analogous to be
basic approach taken to study interstellar medium in galaxies, where also it
is difficult to disentangle the local and non-local effects [31, 32].
Problems with the standard analysis:
However, there are serious problems with the details of the subsequent
procedure adopted by [22]. Hence their conclusion that the scale height is
constant with radius is not correct, this is discussed next. For the above
adopted model, the luminosity density at any point (R,z) in the disk is given
by

L(R, z) = L(0, 0)exp(−R/RD )sech2 (z/z0 ) (2.1)

where z0 is a scale parameter which is an indicator of thickness of disk vertical
distribution at a given radius,R (see [12, 13]; also, see Section 3.1). RD is
the exponential scale length of the radial distribution of surface density in
the disk, and L(0, 0) is the luminosity density at the galactic centre.
Next, the theoretical intensity profile for the above adopted model, say
at a distance R from the galactic centre along the major axis as seen by
the external observer, is obtained by integrating the contributions from the
non-local regions along the line of sight which are at different radial distance
w.r.t. the galactic centre. In a typical real galaxy, z0 would be expected
to vary with the radial distance from the galactic centre, hence integrating
over this contribution is difficult. To simplify this, [22] assume that the scale
height z0 is constant with radius in the disk. With this assumption, it is
straightforward to obtain the net edge-on distribution of surface brightness
or intensity as seen from a point at a distance R from the galactic centre,
by integrating the luminosity density along the line of sight. The resulting
model surface brightness is given as:
 
R R
I(R, z) = I(0, 0) K1 sech2 (z/z0 ) (2.2)
RD RD
where I(0, 0) = 2RD L(0, 0) and K1 is the modified Bessel function of the
second kind.
Then by comparing this model profile with the observed intensity profile
measured at a cut normal to the major axis at a given R, [22] obtain the

12
value of z0 . However, note that the above expression for the the model surface
brightness (Eq.(2.2)) was derived assuming the scale height z0 to be constant
with radius, which is not physically justified.
Next, they stack the intensity profiles taken along normal cuts at differ-
ent R together, at an arbitrarily chosen z ′ point; and claim that these show
a small spread. Hence they conclude that these profiles are well-fit by a
single composite curve with a constant z0 . Therefore they claim that their
assumption of z0 being constant with radius, used to obtain the model inte-
grated surface brightness along the line of sight, (Eq.(2.2)) from (Eq.(2.1)),
is justified. This procedure proposed by [22] is routinely used by others to
determine the scale height for edge-on galaxies (e.g., [33, 34]).
However, there is a serious flaw with the above approach proposed by
[22], as shown next. Their claim that the profiles at different R are well-fit
by a single composite curve with a small spread around it – thus indicating
a constant scale height – itself has been questioned by [35], and [36].
Using the same data and the composite profile as obtained by [22]; [35]
show that, in fact, there is a finite spread in the observed data for surface
brightness around the composite profile. Further, [35] show that since the
spread is given in units of mag arc sec−2 , it can actually mask an increase
in scale height, of as much as a factor of 1.7-2.5 within the optical disk, in
the two galaxies they studied: namely, NGC 891 and NGC 4565. See section
4.2.7 for details of [35]; also, see Section 5 on a flaring stellar disk. Indeed,
flaring is now shown to be a generic feature of a galactic stellar disk.
The same conclusion, namely, that the scale height z0 is not constant with
radius, was reached by [36]. In an early and perceptive paper, [36] pointed
out that the scale parameter z0 will in general depend strongly on R, and a
constant scale height is only possible for some specially chosen variation of
stellar velocity dispersion with R. Hence they investigate variation of z0 with
R in the observed data. They point out that since we see a projected system,
z0 (R) cannot be measured directly; rather, we see a convolved quantity for
the luminosity. Hence for a particular trial function for z0 (R), they match the
resulting model intensity profiles, obtained by directly integrating the density
profile (Eq. (2.1)) along the line of sight – without making an assumption
of constant z0 (as done by [22] to obtain (Eq. (2.2)) – with the observed
intensity profiles.
Further, [36] point out that the use of constant z0 does not give a good fit
to the data, which is in contrast to the claim by [22]. Instead, [36] try some
physically motivated, simple analytical functions for z0 (R); and find that a

13
best-fit between their models and observed profiles from [22] is obtained when
z0 is taken to increase exponentially with radius with a scale length, RD (see
Figs. 1 and 2 in [36]). This was a striking result; but unfortunately, this paper
was not much noticed in the literature. In another early study, [29] showed
from their data on edge-on galaxies – by doing the fitting over two radial bins
– that the stellar scale height increases with radius. Unfortunately, none of
these important papers, showing the result that the scale height of the stellar
disk increases with radius, has received much attention in the literature.
New, correct iterative approach proposed:
To do the problem correctly; namely, to determine the scale height from
the observational data from external edge-on galaxies accurately, the possible
scale height variation with R must be taken into account. To take account of
the cumulative effect of the possible variation in scale height at points along
the line of sight, the fitting has to be done iteratively, to get the optimal
choice of z0 (R) which gives the best-fit to the observed intensity profiles. In
this task, different density profiles, that is, a functional form for ρ(z) other
than sech2 , could also be tried. This is a challenging task, and we hope
this will taken up in future to accurately determine the true scale height, z0 ,
and its possible variation in R. This could also give us some understanding
about the physical process that gives rise to the radial variation in z0 . This
general approach proposed above is logically the next step, long overdue, in
the approach initiated along this direction by [36], who had chosen a specific
trial function of z0 (R).
In an important recent study, [37] model the observed intensity profile
from NGC 7572 as a sum of a thin disk and a thick disk, with the aim to
study the variation of disk thickness with radius. To do this, they use model
intensity profile as a sum over two sech2 profiles, and let the thickness of
each of these vary with R. Despite allowing for this variation, they find that
best-fit case still left strong residues (that is, an excess of emission from the
observed data) at high z compared to their model best-fit profiles. They point
out that this excess indicates contribution by non-local regions, that is, from
different points along the line of sight that cuts the major axis at R. They
note that this excess occurs because the simplified mathematical method used
does not correctly capture the contributions from non-local regions, when
doing the double sech2 fits. They caution that while their results clearly show
the trend of a flaring disk; in view of the approximate mathematical model
used, the exact flaring values cannot be given by their work. We stress that

14
this again points to the necessity of doing an iterative mathematical fitting
of observed intensity profiles, as we have proposed above.
There is motivation for taking up this problem, because theoretical stud-
ies, as well as a different line of observations: namely, the star counts data
for the Milky Way; and other direct observational data for the Milky Way,
as for example, from COBE, show that flaring of a stellar disk is a generic
result. It would be interesting to see whether the correct fitting of observed
profiles as proposed above for external galaxies also shows such a flaring
trend. The results for scale height as a function of radius obtained from an
iterative fitting approach (as proposed above) would be a good check against
the resulting model scale height values from theoretical studies done using
the multi-component disk plus halo model. It is possible to obtain the model
scale height values for some galaxies for which input parameters for this
model are known from observations (see Sections 5.5, 8 for details).
This approach of integrating along a line of sight across the entire galaxy
to obtain the net intensity [22, 36] implicitly assumes that the optical depth
effect is negligible across a line of sight in an edge-on galaxy. However, in
reality, a galaxy is optically thick, at least in the central regions, down to
almost the inclination angle of 600 [38].
It should be noted that the presence of other vertical spatial features
such as warps (m = 1 type modes, see Section 7.2) makes it harder to
observationally determine the galactic plane, and hence the vertical density
distribution including the scale height, especially in the outer parts of the
Milky Way. This analysis has been done, and the existence of flaring of stars
[39, 40, 41], and HI gas [42] has been established from the data.

2.1.2. Analysis of 3-D light distribution

Another approach to extract the detailed density distribution including
the vertical structure of edge-on galaxies is to de-project the 3-D light dis-
tribution of an edge-on galaxy in terms of inverse Abel transform [2], see
for example [34]. This method has been applied by [43, 44, 45]. However,
these papers do not address the possible variation of z0 with R; instead, these
mainly focus on the radial variation of intensity including disk truncation.
The effect of dust extinction has been included in the full 3-D modelling of
galaxies [38, 46, 47, 48]. We point out that to do this problem correctly, these
studies must be redone so as to take account of the possible radial variation
of z0 . For that, the fitting has to be done iteratively to get the variation in
z0 with R that gives the best-fit to the data, and thus extract the correct

15
vertical density distribution in the stellar disk.

2.1.3. Analysis of star counts data in the Galaxy

So far we have discussed observations of external galaxies by photometry
to deduce the vertical density distribution. An alternative method to obtain
the vertical density distribution is possible in the Galaxy; namely, to directly
measure the star counts as a function of z. The advantage of using this
approach is that it does not require any conversion from intensity to mass
density that would involve some assumption about mass-to-light ratio in the
disk (as pointed out, for example, in [49]).
At an earlier time this approach was used by Oort [25] who used the
observed star counts data for K-giants in the solar neighbourhood in the
Galaxy as a constraint to obtain the total mid-plane density or the Oort
limit, to be discussed later in Section 3.
In recent years, there have been great strides in observations of star
counts. These permit the measurement of vertical density distribution of
stars, as a function of type, age etc. over a larger radial range of several kpc:
for example, the work done using 2MASS data by [50]; using SDSS-SEGUE
data by [41]; using SDSS data by [51]; and using LAMOST data by [52].
In principle, the vertical density profile for stars can be obtained from
this data in the Milky Way. However, in practice, it is difficult to determine
the density profile. The results seem to depend on the actual choice of data
cuts taken and the analysis done. For example, the analysis of Gaia data
by [53] shows a sech2 profile. A Bayesian analysis by [54], based on the
SDSS data of several million K and M stars by [51], shows the stellar density
profiles to be steeper than sech2 , and it shows a moderate evidence for sech
and exponential profiles compared to a sech2 profile.
However, despite these caveats in the analysis of star counts data, it
is still useful to compare theoretical model results for density profile with
the observations of number density from star counts (see Section 4.1.1, and
Section 8). The observed data could in fact be used to fine-tune or improve
the theoretical models.
Remarkably, new detailed astrometric and kinematical data for the line-
of-sight velocities for individual stars from SDSS and Gaia are now available
for an external galaxy, namely, the LMC (Large Magellanic Cloud). This
involves about 15 million stars out of the two billion stars in the Clouds.
This information has now been used to model a 3-D mass distribution in the
LMC [55].

16
2.1.4. Stellar velocity dispersion data
An important parameter describing a stellar disk is the typical stellar ve-
locity dispersion. This has been measured in the Milky Way by observations
along the so-called Baade’s window, which is a region of low dust extinction.
This permits kinematics of stars to be measured between 1-17 kpc from the
Galactic centre [56]. This study shows that the radial stellar velocity disper-
sion falls exponentially with radius, with a scale length, Rv = 8.7 kpc.
The velocity dispersion measurement had been done for only a few ex-
ternal galaxies until the 1990’s [57], that too, in the inner regions where
the dispersion is high. This situation has changed for the better in the last
∼ ten years due to the availability of 2-D IFU data on stellar kinematics
of external galaxies: for example, see, the DiskMass survey [58]; CALIFA
(The Calar Alto Legacy Integral Field spectroscopy Area) survey [59, 60];
MaNGA (Mapping Nearby Galaxies at Apache Point Observatory) survey
[61, 62]; SAMI (Sydney-AAO Multi-object Integral Field Spectrograph) sur-
vey [63]; also, see [9] for details of various IFU surveys. However, due to
resolution problems, the stellar velocity dispersion measurements from many
of these are still limited to the inner regions only. Often, a single value for
the dispersion in the effective radius (Re ) is extracted from the data.4
However, CALIFA does have measurements up to 1-2 effective radii and
these are well-sampled, see the detailed work by [60] that gives stellar disper-
sion as a function of radius. Similarly, the data from the various surveys also
need to be analyzed to obtain the stellar velocity dispersion values as a func-
tion of radius. These are needed as input parameters to study disk vertical
structure. This will be discussed in detail in Section 8. Given the difficulty
of measuring the velocity dispersion, an alternate way has been suggested
by [64] where the measured line-of-sight velocities are used to estimate the
radial velocity dispersion using the concept of asymmetric drift. Here, the
asymmetric drift is used as a proxy for stellar velocity dispersion (see Fig. 9
from [64]). However, in the above work, the sampling points are few, and the
analysis is limited to the central two effective radii, so the results for velocity
dispersion obtained are not very useful for studies of vertical structure across
the disk.

4
An effective radius, Re , of a galaxy is the radius from within which half the luminosity
of the galaxy is emitted. For an exponential disk, Re ∼ 1.7RD . For a typical galaxy with
a bulge, Re would be smaller than this.

17
 For the Galaxy, using Gaia data, very detailed information – such as the
dispersion versus stellar age – is now available (e.g.,[65]). For convenience,
typically in the literature, starting from [12, 13], as also in this review, all
the stars in a galactic disk are taken to form a single component with the
same velocity dispersion (see Section 4.2.1 for a discussion on this). Recent
data gives the velocity dispersion to be non-isothermal. This information is
used to build a more general theoretical model (Section 4.4.2).

2.2. Interstellar gas: Surface density and velocity dispersion

Although this review mainly focuses on the stellar disk, it also shows that
the interstellar gas needs to included for treating a realistic, multi-component
galactic disk. Hence, we briefly describe here the properties of the interstellar
gas in a typical galaxy like the Milky Way. The interstellar gas consists of
two main mass components: the interstellar atomic hydrogen gas (HI), and
the interstellar molecular hydrogen gas (H2 ). Each of these is distributed in
a thin disk, and is characterized by different values of surface density and
velocity dispersion, which vary radially. The interstellar medium (ISM) in
galaxies is a vast, well-developed field [31, 32]. The observational properties
of ISM are complex and have been extensively studied: for the HI gas, for
early in-depth reviews, see [66, 67]; and for more recent reviews see [68, 69].
For the H2 gas in the Milky Way, see the reviews by [70, 71]; and for reviews
for H2 in external galaxies, see [72, 73]. Thus, the properties of HI and
H2 gas, and their radial distribution, have been extensively studied for the
Galaxy as well as for external galaxies.
The interstellar gas is observed to be distributed in a thinner layer than
the stars, with the H2 distribution being thinner than the HI distribution.
The three disk components are taken to be co-spatial, concentric and copla-
nar; also, see Section 4.1. The ISM is a complex, multi-phase system [74, 75],
consisting of CM (cold medium), WNM (Warm Neutral Medium) and HIM
(Hot Ionized Medium). Despite the multi-phase nature of HI, for simplicity,
we consider it to be characterized by a single value of surface density, Σ(R)
and σHI (R), as is usually done in the literature. This represents the CM
and WNM components of HI [75, 76]. The contribution of HIM to the mass
in the thin disk region is small, hence it is neglected for the problem of the
vertical structure of the disk.
For the purpose of this review we need to consider the observed values of
the global properties such as the surface density and the velocity dispersion
of the HI and H2 components. In the multi-component disk plus halo model

18
to be discussed extensively in Section 4 onward, both HI and H2 gas are
treated as gravitating disk components, which interact gravitationally with
stars and each other (see Section 4.1). The gas is supported by turbulent
pressure and the corresponding velocity dominates over the thermal velocity
in each component [66, 70, 32]. Typically, in the inner disk, the stellar surface
density is about 10 times higher than the gas surface density; and the stellar
velocity dispersion is higher by a factor of few than the HI dispersion, which
in turn is slightly higher than the H2 dispersion. The total mass content in
HI and H2 is nearly comparable: most of the H2 gas is located inside of R=
8.5 kpc, whereas most of the HI gas is distributed in the outer Galaxy [70].
Thus the radial distribution of HI and H2 components is different. The HI
velocity dispersion is supported by turbulence; the pressure due magnetic
fields and cosmic rays is not important for the internal support of HI gas
[77].
The HI dispersion values decrease with radius and then taper off to a
saturation value (e.g., see the discussion in [78]; also, Section 4.2.3). At
the solar position, the observed radial stellar velocity dispersion is ∼ 18 km
s−1 and decreases exponentially with radius; while the isotropic HI velocity
dispersion is ∼ 8 km s−1 which tapers off at large radii to ∼ 7 km s−1
(see Section 4.2.3). The specific values of these parameters chosen, with
some variation as per the galaxy under consideration, will be introduced as
necessary during the application of theoretical models in the later Sections
(e.g., Section 4, Section 6).
In contrast to the optical studies of stars where dust extinction is a prob-
lem; for HI gas – using the 21 cm radio observations as a tracer – the 3-D
HI data cube allows one to get detailed maps of HI distribution. Here the
effect of optical depth is not important since one sees local HI (at a partic-
ular line-of-sight velocity bin) at each point along the line of sight. So it is
possible to obtain a 3-D distribution map of HI in the Galactic disk, as well
as in external galaxies. As an example of later, see the 3-D HI data cube
used to study rotation curves in galaxies (e.g., [79, 80]). This point and the
applications of 3-D HI data cube will be discussed further in Sections 4.5.2
and 6.5.

3. Self-consistent vertical density distribution in a galactic disk

In this section, we develop the topic of self-consistent vertical density
distribution in a galactic disk systematically. We start from the basic, classic

19
model of a thin, one-component, isothermal disk; to be followed in Section 4
by recent, more realistic multi-component disk plus halo model.
At the outset, we point out that at a basic level the vertical structure is
more difficult to study than the planar mass distribution in a rotationally
supported galactic disk. A galactic disk is mainly supported by rotation in
the plane; the random velocities do not play a major role in its support.
For a disk in rotational equilibrium, the rotation velocity at any point is
determined by the global or non-local mass distribution in the disk. The
motion of a particle in the plane can be treated as a test particle in the field
of this disk. One can then use the observed rotation curve as a constraint
to deduce the radial mass distribution in a galaxy. A galactic mass model
typically is built on this idea (e.g., [81, 82, 83]. This has in fact been used to
show the important result of existence of dark matter (e.g. [84, 85]; also, see
Section 6), although the details of the deduced mass distribution are more
complicated and depend on the actual geometry [2].
On the other hand, for a thin disk in hydrostatic equilibrium, the vertical
motion is determined locally at any radius. This is because the vertical gradi-
ent of the gravitational force is dominant and hence the vertical density distri-
bution is determined by the local conditions, as shown later in this section.
Thus, interestingly, the disk vertical structure can to be used to uniquely
trace the local gravitational potential at a given radius; this is in contrast to
the planar case. However, the entire vertical density distribution along the z
direction at a given radius has to be determined self-consistently by jointly
solving the Poisson equation and the equation of hydrostatic equilibrium. In
a limiting case, at low z values close to the mid-plane – corresponding to
the region of nearly constant density; the motion of a single particle can be
treated as that of a test particle, and it follows a simple harmonic motion
[86, 2]. However, that does not permit us to obtain the entire ρ(z) profile.
In contrast, in the planar case, using the test particle approach one can use
the observed rotation curve to build the entire galaxy mass model. This is
what makes the vertical structure of a galactic disk, or the mass distribution
along z, harder to study than the planar case.

3.1. Basic, classical model of a galactic disk

The vertical density distribution of a galactic disk is typically obtained by
treating it as a self-gravitating, isothermal system supported vertically by its
internal pressure gradient. In a pioneering work, Spitzer (1942) [12] showed

20
that the vertical density distribution of such an idealized disk has a sech2
form, as discussed next. We discuss this giving the details of the physics in
the derivation. As for any self-gravitating system, the vertical structure and
dynamics of a galactic disk are directly related through the Poisson equation.
For an axisymmetric galactic disk, the ϕ component of the Laplacian is zero.
[12] considered the disk to be locally dense, or that the local density of
matter in the disk is much greater than the the average density of the rest of
the galaxy interior to that point R; hence the effect of the rest of the galaxy is
neglected. Thus, to a first approximation, the disk distribution is treated as
a one-dimensional problem along z. Thus, in [12] the R term in the Poisson
equation is neglected and the Poisson equation is taken to reduce to:

∂ 2 Φ(R, z)
= 4πGρ(R, z) (3.1)
∂z 2
where both Φ(R, z), the gravitational potential and ρ(R, z), the mass density
are given at a local radius R. Thus, the force along z, and the vertical density
distribution, ρ(z), are determined locally, and depend only on the physical
conditions at a given radius R. Thus, both are taken to vary purely as a
function of z.
The disk is assumed to be supported against gravitational collapse along
the z direction by the vertical gradient of its own internal pressure. That
is, the self-gravitational force is balanced by the gradient of the pressure
along the vertical direction: this decides the equilibrium vertical density
distribution. Typically, the pressure support is taken to be due to the velocity
dispersion along z, hence the pressure is given by P = ρ(z)⟨(vz2 )⟩ where ⟨(vz2 )⟩
is the mean square velocity dispersion along z. The disk is taken to be
isothermal along z for simplicity, so that the velocity dispersion is constant
along z.
The force equation along the z-axis, that is, the equation of hydrostatic
balance, is given by (e.g., [87]):

⟨(vz2 )⟩ ∂ρ
= Kz (3.2)
ρ ∂z
where Kz = −∂Φ/∂z is the gravitational force force per unit mass along the
z-axis. For a reasonable physical behaviour of pressure falling with z, the
assumption of hydrostatic balance (Eq. 3.2) is satisfied in a typical galactic
disk.

21
 The distribution is taken to be stationary or steady-state with no mass
motions along the z direction, hence the continuity equation is not employed
to obtain the vertical density distribution.
The self-consistent vertical density distribution, ρ(z), is obtained by solv-
ing the Poisson equation (eq. (3.1)) and the equation of hydrostatic equilib-
rium (Eq. (3.2)) together. On combining the above two first-order, linear
differential equations, the following second-order, linear differential equation
in ρ(z) is obtained. This is the joint Poisson-hydrostatic balance equation
for a one-component disk:
 2 
∂ ⟨(vz )⟩ ∂ρ
= −4πGρ (3.3)
∂z ρ ∂z
The solution for this equation was obtained by [12] to be:

ρ(z) = ρ0 sech2 (z/z0 ) (3.4)

where ρ0 is the mid-plane density and the parameter z0 is the scale param-
eter which is a measure of disk thickness or scale height. (See Appendix
A for a detailed discussion of vertical scale height for a gravitating disk.)
This solution is obtained analytically in terms of a subsidiary equation as
explained in detail in [87]. For small z << z0 , the above solution for the
density, ρ(z) (Eq. (3.4)) reduces to: ρ(z) = ρ0 exp − (z/z0 )2 ; that is, a Gaus-
sian form. At the other limit of z >> z0 , the above solution reduces to:
ρ(z) = 4 ρ0 exp(−2z/z0 ); that is, an exponential.
The two boundary conditions used to solve the above second-order differ-
ential equation (Eq. 3.4) are: first, the disk surface density, Σ, at the region
is constant (as discussed later in this section); and second, dρ/dz=0 at the
mid-plane, or z = 0 [87]. The latter condition arises because for a realistic
distribution, the density is homogeneous close to the mid-plane. These two
conditions are used to determine the two parameters ρ0 and z0 .
The resulting values of ρ0 and z0 , are specified in terms of physical quan-
tities, namely the surface density and the velocity dispersion, and are given
as follows:
1/2
2πGΣ2 ⟨(vz2 )⟩ ⟨(vz2 )⟩

ρ0 = ; z0 = Σ/ρ0 = = (3.5)
⟨(vz2 )⟩ 2πGΣ 2πGρ0
On substituting ρ(z) (from Eq. (3.4)) into Eq.(3.2) gives the expression for
|Kz |, the vertical force per unit mass, as follows:

22
 ⟨(vz2 )⟩ z
|Kz | = 2 tanh (3.6)
z0 z0
Physically, the scale parameter z0 , a measure of disk thickness, is set by
the equilibrium between the local gravitational force and the internal pressure
gradient as given by the equation of hydrostatic balance.
We will refer to this as the sech2 model in the rest of the review article.
Recall that the assumption given above (in [12]), namely, that the disk is
locally dense, led to Eq. (3.1). Hence the resulting density distribution,
ρ(z), depends only on the local conditions in the disk. This classic solution
for ρ(z), the vertical density distribution along z (Eq. (3.4)), is taken to be
the standard model and routinely applied to describe the vertical structure
of a galactic disk, as discussed below.
The vertical structure of an isothermal, self-consistent distribution was
also studied by [13] who assumed a disk of constant surface density so that
there are no radial gradients in density or force. Hence, the radial component
of the Poisson equation is neglected and the problem again reduces to a one-
dimensional problem of distribution along z. [13] also obtained the same
solution for the vertical density distribution as given by Eq. (3.4).
Historically, it was later shown quantitatively and more rigorously by
Oort and others that the above simplified form of Poisson equation that
includes only the z term in the Laplacian (Eq.(3.1)), is valid in the solar
neighbourhood for the Milky Way. To show this, the R term of the Poisson
equation was obtained in terms of the observed rotation curve and its local
derivatives, and was shown to be less than (4πGρ), and hence the R term
could be neglected [88, 5, 86, 89]. The same result was shown to be true using
modern data – for rotation curve parameters and the Galactic constants and
the mid-plane total density – at the solar neighbourhood, as well as at the
other radii in the inner Galaxy [90].
The above simplified form of Poisson equation as given by Eq.(3.1) has
been shown to be valid on general physical grounds for a typical thin disk as
well (see Section 2.2 in [2]). Thus, the Poisson equation for the thin disk also
contains only the z term, so it has the same simplified form as used by [12]
for a locally dense disk. Hence, the density distribution is a one-dimensional
problem along z for a thin disk as well. The vertical force, and the resulting
vertical density distribution, for the thin disk are determined locally, and
depend only on the local surface density at a given point in the disk. Thus
the condition of high local density as assumed by [12]; and the condition

23
of a thin disk, are equivalent and give rise to the same formulation for the
vertical structure problem. Eq. (3.1) shows that the planar and vertical
dynamics are decoupled for a thin disk, e.g., [2], as well as for a locally dense
disk considered in [12]. We underline an important point that, for the one-
dimensional, local problem considered here; the surface density is constant at
a given R. Hence, the use of local surface density being constant as one of the
boundary conditions used in solving Eq. (3.3) is justified. The hydrostatic
balance equation remains the same in both approaches. Thus, the resulting
density distribution for a thin disk is also specified by Eq.(3.4).5
It is worth pointing out that in the literature, the paper by Spitzer [12],
and the resulting sech2 density profile (Eq.(3.4)) from that work, are rou-
tinely cited in the literature in the context of a thin disk. However, [12] did
not explicitly mention the assumption of a thin disk in that paper; rather,
the paper was formulated for a locally dense disk. As discussed above, the
treatment and the results for ρ(z) are identical for both these assumptions.
Hence the use of the classic sech2 solution [12] (obtained for a locally dense
disk), to also formally denote the self-consistent vertical density distribution
for a thin, stars-alone, one-component galactic disk – as is normally done the
literature, is justified.
Strictly speaking, while obtaining the terms involving the z derivative of
the potential while deriving the Poisson equation (Eq.(3.1)), and the equation
of hydrostatic balance (Eq.(3.2)); the complete Jeans equation along z (e.g.,
Eq. (4.29 c) from [2]) should be used. This would also include the cross
terms and terms involving kinematic effects. This is not generally noted in
the literature and the above simplified forms are routinely used in subsequent
papers on the topic. It has been shown that these terms are small for a
typical, thin disk (e.g., [2]), as explained in detail next.
The complete Jeans equation along z (Eq. 4.29 c from [2]) also includes
terms containing the cross term < vR vz >, where the average is taken over
the velocity dispersions. The term < vR vz > is a component of the velocity
ellipsoidal tensor whose value is set by the tilt of the velocity ellipsoid w.r.t.
the galactic plane. For any general orientation of the velocity ellipsoid that
is pointing toward the galactic center, it can be shown that the ratio of the

5
Although Eq.(3.1) was shown to be valid for a thin disk by [88, 2], they use this
formulation to study the mid-plane density – but do not obtain the solution ρ(z) for a
thin disk.

24
two cross terms involving < vR vz > (which are dropped), to the other terms
in Eq. (3.2) (which are kept), is ∼ z 2 /(RRD ) [86, 2]. This ratio is small
and hence these terms can be neglected for the thin disk in general [2]. For
example, this ratio is < 0.01 for z < 1 kpc for the solar neighbourhood (e.g.,
[89, 91]). Hence the above simplified form for the z term of Poisson equation
(eq.(3.1)) and the equation of hydrostatic balance (Eq. (3.2)), as is generally
done in the literature, is justified for a typical, thin galactic disk.
Similarly, the cross terms were shown to be negligible for a thin disk,
while evaluating the R-term of the Poisson equation which was then shown
to be small compared to the z term for a thin disk, see [5, 86, 89]. The cross
terms being small is the reason why the R-term could be evaluated simply
in terms of the observed rotation curve and its derivatives, as done in these
papers. The significance of the cross terms in the context of a thick or low
density disk will be discussed later in Section 4.3.1 (in particular, see the
discussion after Eq. (4.9) and Eq. (4.11).
The above solution for the vertical density distribution (Eq.(3.4)) has
been the workhorse of vertical disk structure and is taken as the default
model for the vertical disk structure. However, it was found that it does not
explain the recent observed data. This was one motivation for the formulation
of a more realistic multi-component galactic disk plus halo model (Section
4).

3.1.1. A brief comment about the Oort limit

At around the same time as the above model was proposed; a parallel,
independent line of study by Oort(1932) [88] focused on the determination
of the mid-plane density in the solar neighbourhood, which is now known
as the Oort limit in his honour. Here, the same two equations (the Poisson
equation and the hydrostatic balance equation along z) are used; but then
the observed values of number density of stars obtained by star counts are
used to determine the force Kz and hence the mid-plane density.
If the disk consists of several sub-components, then the equation of hy-
drostatic balance for a particular population of tracer stars has the following
form, where its pressure gradient is balanced by the net gravitational force
of the entire disk:

⟨(vz2 )i ⟩ ∂νi
= Kz (3.7)
νi ∂z
Here νi (z) is the number density of tracer stars (denoted by the subscript

25
i), at a given height z from the mid-plane; and < (vz2 ) >i is the velocity
dispersion of these tracer stars, and it is assumed to be isothermal. Kz , the
gravitational force per unit mass, is due to the entire disk. The Poisson
equation for the entire disk has the same form as for a one-component disk
(Eq.(3.1) where Φ now denotes the total potential of the disk.
Thus, combining Eq.(3.7) and Eq.(3.1), the joint Poisson-hydrostatic bal-
ance equation that describes the vertical density distribution of the particular
tracer component is given by:
 2 
∂ ⟨(vz )i ⟩ ∂νi
= −4πGρ (3.8)
∂z νi ∂z
By measuring the observed number density of a particular population of
stars used as tracers and using the above equation; one can, in principle,
estimate the mid-plane density. It should be pointed out that such an esti-
mate of the mid-plane density would be very uncertain because it involves
a triple differentiation of star counts: once to get the number density as a
function of z, and twice more (as seen from the l.h.s. of the equation of joint
Poisson-hydrostatic balance equation (Eq. (3.8)), see [2] for details.
However, the uncertainty inherent in this method can be reduced by using
different tracer stars to obtain the mid-plane density, and taking the average.
This was done using F-stars and K-giants as two different set of tracers by
Oort(1932) [88], (also, Oort (1960), [25]), who concluded that ρ0 , the total
mid-plane density, in the solar neighbourhood is equal to 0.15 M⊙ pc−2 . This
is now called the ”Oort limit”. This pioneering work attempted to measure
the total mid-plane density in the solar neighbourhood. By comparing with
the observed density of gas and stars at that time, Oort concluded that
∼ 40% of this total mass density is not accounted for, or is ”missing”. It
was conjectured at that time that this ”missing” or unseen mass could be in
the form of faint white dwarfs or interstellar gas clouds [86]. Later this came
to be known as the local dark matter, see, e.g., [92], for a discussion of the
possible constituents of this local dark matter and the early searches for it.
Later, the Oort limit problem was reworked by [93] who aimed to reduce
the uncertainty introduced by multiple differentiation of observed data. To
do this, [93] considered the disk to consist of 16 different isothermal stellar
components, and following a somewhat different procedure obtained the mid-
plane density, ρ0 to be ∼ 0.18M⊙ pc−3 , see [2] for details. At that time, the
observed total mass density in the solar neighbourhood, including the stellar
and gas density, and estimates of the density in white dwarfs as relics of

26
stellar evolution, was found to be be ∼ 0.114M⊙ pc−3 . Hence the missing
mass or the local dark matter density was deduced to be ∼ 0.066M⊙ pc−3 or
36.7 % of the total mid-plane density.
This problem has been extensively worked on over the years, using more
modern data: for example, using the HIPPARCOS data (from the ESA
HIPPARCOS satellite)[94], [95]; using RAVE (RAdial Velocity Experiment
survey) data for red clump stars [96]; and using the HST data for M dwarfs
[97]. The analysis to determine the total mid-plane density, and the resulting
value for it, have been heavily debated in the literature: see for example, [98]
and [99]. More recently, the local total surface density as well as the local
dark matter density has been extensively studied: using Gaia DR2 and/or
LAMOST data [100, 101, 102]; and using APOGEE and Gaia data [103].
For some recent reviews on the local dark matter density, see [104, 105].
This section gives a brief introduction, including the early history, of this
important topic. The determination of the Oort limit, and the contribution
of the observed or baryonic components to it, continues to be a subject of
active research. The estimate of the total mid-plane density, and hence local
dark matter density, is of great significance for the study of galactic dynamics.
This is because it gives a local constraint on the large-scale Galactic mass
model, as well as the extended dark matter halo distribution. The total
local mid-plane density has interesting implications for the formation and
evolution of disk galaxies [94].
Extensive efforts have been made to study the nature and constituents
of the local dark matter (e.g., [92, 106]); but no clear, convincing candidate
has emerged so far.

3.2. Gas in a galactic disk in the classical treatment

The interstellar gas typically contains a small mass fraction (∼ 10 −
15%) of the disk mass (e.g., [72, 107]). Hence, typically, in the early work,
the self-gravity of the gas was ignored even when obtaining its own vertical
distribution. Instead the gas was treated to consist of massless test particles
responding to the dominant gravitational potential, namely that of stars.
The gas distribution, ρg (z) is then shown to have the following Gaussian
form [87]:

ρgas (z) = ρs0 exp − (z/z′0 )2 (3.9)

where z0′ = σg2 /(2πGΣs ), and ρs0 is the total mid-plane stellar density in
the disk, and σg is the gas velocity dispersion. Note that the above result

27
for ρgas (z) depends only on the stellar gravitational potential, and the gas
velocity dispersion; and has a Gaussian form. This form agreed fairly well
with basic form of the early observations [108, 75, 67]. This has been the
standard usage in the literature for the vertical distribution of interstellar
gas in a galactic disk, for convenience. However, it does not take account of
the gas gravity, which is shown to affect the gas distribution (see Section 4,
and Appendix A).
In this scenario, the atomic hydrogen gas, in presence of the stellar disk
potential alone, should have a scale height which increases exponentially
with radius, since the stellar surface density falls exponentially with radius
while the gas dispersion is nearly constant with radius (see Section 4.2.3, and
Section 6.5). However, what is observed is the opposite in the inner Galaxy.
In fact, the observed near-constancy of HI scale height in the inner Galaxy
(< 8 kpc) had been a long-standing puzzle Oort(1962) [109, 75, 67]. This
led [16] to propose that the gas gravity should be included since it would
decrease the gas scale height, which could better explain the observed gas
thickness. This was indeed one of the motivations for the multi-component
disk model proposed by [16], as will be discussed in detail in the next section.
As discussed in that section, the gas gravity needs to be taken into account
to get the correct physical description of the vertical structure of all the disk
components, including gas.
The flaring of HI gas is in fact observed and well-established at larger
radii, in the outer Galaxy [42, 67, 110, 111], also see Section 4.2.6, and
Section 6. The physical origin of the gas flaring seen at large radii can be
explained naturally by the multi-component disk plus halo model, see Section
4.2.6, and Section 5.

3.3. Vertical constraining effect of molecular cloud complex on disk

In contrast to the classical treatment which ignores the gas gravity; the
vertical constraining effect due to the gravitational field of a typical molecular
cloud complex on the the galactic disk stars was studied in a remarkably
prescient paper by [112]. They noted that nearly half the interstellar gas in
the Milky Way is in a molecular hydrogen form, and ∼ 90% of it is in the
inner Galaxy [70]. This gas is contained in giant molecular clouds; which
are further segregated into well-defined, large, discrete gravitating features
which have been variously called complexes or clusters or chains, as shown
in the early work [113, 114, 115]. The typical molecular cloud complex has

28
a size of a few 100 pc, has an oblate spheroidal shape with the full height of
120 pc, and a mass of ∼ 107 M⊙ each. Such massive complexes are seen in
other galaxies as well (see [112] for details).
[112] pointed out that the average observed density of H2 in a typical
complex is ∼ 16cm−3 or ∼ 1M⊙ pc−3 . This is about six times higher than
the dynamical or total mid-plane mass density in the solar neighbourhood,
or, the Oort limit (Section 3.1.1). Such a dense, massive cloud complex
would dominate the local gravitational field in a galactic disk. Motivated
by this, [112] studied the dynamical effect of such a complex on the vertical
distribution of stars. To do this, the mass distribution in a cloud complex
is taken to be that of an oblate spheroid of constant density. The force per
unit mass along z due to a complex is taken from [116], who had calculated
it in the context of galaxy mass models.
To get an idea about the effect of the complex, the gravitational force
per unit mass,|Kz |, due to a stars-alone disk (Eq.3.4); and that due to the
complex at the centre of the complex (r = 0) vs. z are plotted in Fig. 2.
Here r is the radial distance measured from the centre of the complex in the
disk plane.6 The centre of the complex is located at R = 8.5 kpc. It can
be seen that for z < 200 pc, the force due to the complex dominates over
that due to the disc; with the ratio being a maximum, equal to 9.5, at the
outer edge of the complex (z = 60 pc). A similar result is seen over the
radial extent of the complex (see Fig. 2 in [112]). Thus, the force due to
the complex dominates over the self-gravity of the stellar disk over a large
spatial range in R and z. This has a strong effect in modifying the stellar
distribution near the mid-plane, as shown next.
The equations to be solved to obtain the modified stellar mass distribution
in presence of a complex are: the equation of hydrostatic balance, or the force
equation along the z direction:

⟨(vz2 )⟩ ∂ρs
= (Kz )s + (Kz )complex (3.10)
ρs ∂z
and, the joint Poisson equation for the stars and the complex:

6
The parameter r is used later to denote the radius in the spherical polar co-ordinates,
as in the usual notation, in Section 4.3.2

29
Figure 2: Plot of |Kz |, the vertical gravitational force per unit mass due to a cloud complex,
and that due to the undisturbed stellar disc, versus z, the distance from the mid-plane,
at the complex centre. For z < 200 pc, the force due to the complex dominates over that
due to the disk. The ratio is a maximum, equal to 9.5, at the outer edge of the complex,
at z = 60 pc. Source: Taken from [112]

∂(Kz )s ∂(Kz )complex ∂(Kr )complex (Kr )complex

+ + + = −4πG(ρs + ρcomplex )
∂z ∂z ∂r r
(3.11)
The complex being massive is assumed to be unaffected by the disk, hence
the joint Poisson equation effectively reduces to that for the redistributed
stellar case alone:

∂(Kz )s
= −4πGρs (3.12)
∂z
where ρs denotes the redistributed stellar density in the field of the complex.
On combining the above two coupled first-order linear differential equations
(Eq. (3.10) and Eq. (3.12) ), gives the following second-order, linear differ-
ential equation in ρs :
 
2 ∂ 1 ∂ρs ∂(Kz )complex
< (vz )s > = −4πGρs + (3.13)
∂z ρs ∂z ∂z

30
Figure 3: Plot of self-consistent density for self-gravitating, undisturbed stellar disk
(dashed line), and the modified stellar disk density in the field of a cloud complex (solid
line) vs. z, at at the centre of the cloud complex. The complex centre is located at R=8.5
kpc. The modified mid-plane density is 2.6 times higher and the vertical scale height
(HWHM) is 3.4 times smaller than the undisturbed case, showing the strong constraining
or pinching effect due to the cloud complex. Source: Taken from [112]

This equation is solved numerically as an initial-value problem from the

inside-out, using the fourth-order Runge-Kutta method [117]. The initial
condition of dρs /dz = 0 at z=0 is used since that would be true for any
realistic mass distribution in a disk. The value of the redistributed density
ρs at the mid-plane (z = 0) is not known, a priori. Its value is obtained by
trial and error, by assuming that the total column density of the undisturbed
stellar disk, Σs , remains constant even in the presence of a complex. This is
justified because the complex is assumed to redistribute the stars only along
z at a given radius, since the radial force due to the complex can be neglected
as argued by [112]. Hence, the determination of density distribution for stars
in presence of a complex becomes effectively a one-dimensional problem along
z. To illustrate the effect of the gravitational field of the complex, consider
the vertical stellar density distribution at the central position of the complex
(r=0), located at R=8.5 kpc. On solving Eq. (3.13) for (kz )complex =0, the
undisturbed sech2 distribution [12] is obtained, as expected (Fig. 3, dashed
line); while including the effect of complex results in the modified stellar
distribution (Fig. 3, solid line). The gravitational force due to the complex

31
Figure 4: Plot of vertical scale height (HWHM) for the modified stellar distribution for
stars in the field of the cloud complex vs. r, the radial distance from the centre of the cloud
complex (solid line); and the constant scale height for the undisturbed stellar disk (dashed
line). A striking result is that the vertical constraining by the complex is significant over a
large radial distance ∼ 500 pc from its centre; this is due to the extended mass distribution
within a massive complex. Source: Taken from [112]

strongly alters the vertical distribution of the stellar disk: the mid-plane
density increases by a factor of 2.6 to be equal to 0.19M⊙ pc−3 , while the
scale height (HWHM) decreases by a factor of 3.4 to be equal to 88 pc.
Thus, overall the density distribution is constrained to be closer to the mid-
plane. This strong ”constraining” or ”pinching” effect is a generic result,
arising due to the additional gravitational force of the complex acting on the
stellar disk.
Repeating this procedure at different radial distances, r, from the centre
of the complex (in the galactic plane) gives the resulting modified vertical
scale height (HWHM) at each r. This is plotted as a function of r, (see Fig.
4, solid line); also plotted is the constant scale height for the undisturbed
stars-alone case (Fig. 4, dashed line). The HWHM of the stellar distribution
decreases by a substantial factor of ∼ 3.4, at r = 0 or the central location
of the complex, as already shown in Fig. 3. Another striking result is that
the trough representing the scale height decrease is broad; so that the scale
height reverts back asymptotically to its original undisturbed value of 299
pc only beyond 1 kpc (not shown in Fig. 4). Thus the resulting scale height

32
distribution of stars is locally corrugated. The vertical distribution of HI
gas in the galactic disk also shows a similar constraining effect due to the
gravitational field of the complex.
Thus, a typical dense, massive and extended molecular cloud complex
in a galactic disk extends a considerable constraining effect on the vertical
distribution of stars and HI gas for a sizeable distance (∼ 500 pc) within
its neighbourhood (see Fig. 4). This result suggested that the constraining
effect due to the gravitational force of the average gas distribution in the
disk, and the dark matter halo, on the stellar disk could also be important.
This was the theoretical motivation for proposing the multi-component disk
plus halo model [16], which will be discussed extensively in the next section.
Due to the superposition of effect of several such complexes, the gravita-
tional potential in the disk would be distinctly non-uniform. This would be
of interest for general galactic disk dynamics, such as for heating of stars (see
[112] for details). Such complexes are seen in other galaxies as well. Hence,
this topic needs to be explored further.
Recently, [118], analyzed the high resolution archival HI data in the
Galaxy from the HI4PI Collaboration [119]; and clearly detected this con-
straining effect due to a molecular cloud complex in the Galaxy. This effect
has been detected around the molecular cloud complex region W43-W44.
This complex is located serendipitously near the tangent point region be-
tween the Galactic longitude of ∼ 200 − 400 ; so that the effect of the tangent
point distance ambiguity is minimal. In a real galaxy, one does not see the
effect of an isolated complex. Rather, the constraining effect of a set of cloud
complexes of different masses, located at different locations (R, ϕ), would
be seen in superposition. This makes the analysis and the identification of
the constraining effect due to a single complex (as predicted theoretically by
[112]; see Fig. 4 in this Section) more challenging. However, this detection
has been achieved by taking cuts in velocity space along a given line of sight
(see [118] for details).

4. Multi-component disk plus halo model

In this section, we describe the modern theoretical study of the vertical
structure of a realistic galactic disk. The disk is treated as a gravitationally
coupled multi-component disk system in the gravitational field of the dark
matter halo. Each disk component is assumed to be isothermal with a dif-
ferent velocity dispersion. We first consider the simple case of a thin disk

33
(Section 4.2); followed by the thick or a low density disk case (Section 4.3).
Later, we consider a more general treatment that includes various kinematic
effects, and the effect of non-isothermal velocity dispersion (Section 4.4). The
aim is to start with simplifying, physically motivated assumptions such as
that of a thin disk; and then to consider progressively more detailed physics
in the problem, and thus to study more realistic cases. These cases were
motivated by possible applications to different regions and types of galaxies.

4.1. Motivation and basic outline of the model

4.1.1. Motivation
The initial studies [22, 28] showed that the intensity profiles (I vs z) did
overall agree with a sech2 distribution expected for a self-gravitating, isother-
mal disk ([12], Section 3.1). However, the I-band and R-band observations in
1980’s began to reveal that the observed intensity profiles (e.g., [14, 15]) are
generally steeper, with an excess near the mid-plane; compared to the sech2
distribution that is expected for an isothermal, single-component disk. The
observed profiles could instead be better fitted by either a sech or an exponen-
tial function. A phenomenological model for an isothermal, one-component
disk was proposed by [14] to explain this, but it was somewhat ad hoc and
did not have a clear physical basis.
In fact, since a thin, isothermal, single-component stellar disk can only
result in a sech2 profile (Eq. (3.4)); it is obvious that some additional physics
has to be invoked in the theoretical analysis to explain the observed steeper
profiles. This point has not been clearly recognized in the literature. It
turns out that the theoretical model proposed on physical grounds [16], as
discussed next, naturally results in steeper profiles. This will be discussed in
detail in this section.
To theoretically understand the disk vertical structure, a physically mo-
tivated model was proposed and developed by Narayan & Jog (2002) [16].
This model considers the vertical structure of a realistic multi-component
galactic disk which includes gas gravity; and treats the stars and gas on an
equal footing. The stars and gas are gravitationally coupled, and the disk
is taken to be in the gravitational field of the dark matter halo. The aim
of the model is to obtain the self-consistent vertical density distribution of
each disk component in terms of the physical parameters, such as the sur-
face density and velocity dispersion of the disk components. The additional
gravitational force between the coupled disk components, and the effect of

34
the halo, naturally results in a vertically constrained, and hence a steeper
density profile for each disk component, as we will discuss in this section.
The idea for the above model was suggested from the previous theoretical
work by [112]. That work showed that a typical molecular cloud complex in
the Milky way significantly constrains the vertical stellar distribution around
it (see Section 3.3). It was then natural to ask how a uniform distribution of
co-spatial stars and gas that interact gravitationally would affect the vertical
distribution of each other. The other motivation to include the gas self-
gravity was to see if this could explain the puzzling near-constancy of gas
scale height observed in the inner Galaxy, which could not be explained by
treating the gas as massless test particles responding to the stellar potential
as done in the past (see Section 3.2).
The work by [16] is the first fully self-consistent treatment in the literature
to study the disk vertical structure by considering a gravitationally coupled
disk in the field of the dark matter halo. This model aims to give a physically
complete picture involving all the disk components and the dark matter halo;
and the treatment is rigorous. In the past, [89] had formulated the problem
in a self-consistent fashion but the focus of the work was not on solving
for the self-consistent density profile as shown here. Instead, the observed
stellar density distribution (in terms of star counts data) was used as input
to deduce the total mid-plane density, or the Oort limit (see Section 3.1.1 for
details). Further, the halo was used as a perturbation in that work [89]. The
importance of including gas gravity to study the gas vertical structure was
pointed out by [14, 120]; but the effect of gas on stars was not considered,
and the problem was not solved in a self-consistent way.
The multi-component disk plus halo model proposed by [16] was subse-
quently investigated in a systematic way over the years in a series of papers.
These include various detailed physics points – as mentioned in the first
paragraph in this Section. These cases will be reviewed methodically in this
section. Indeed, the focus of this review is to present this model representing
a realistic galactic disk and the results from it.
As we will discuss, this model explains quite well the observed disk thick-
ness for stars and gas in the Milky Way; and the modern observational trends.
These include a departure from the sech2 law, indicating a steeper profile; and
wings at high z. These results arise naturally from this physically motivated
model [16, 17, 19]. Further, the results for ρ(z) from this model can now be
directly compared with the observed data for star counts in the Milky Way
(see Section 8 for details). Interestingly, in a different but complementary

35
approach, the results from this model could also be directly compared with
the density profile values, ρ(z), obtained from N-body simulations.

4.1.2. Basic outline of the model

The following treatment for the multi-component disk plus dark matter
halo model is from [16], who first formulated and studied this approach in
detail. The treatment is general and valid for any galactic disk, although
this was applied by them in the context of the Milky Way. The results from
this model obtained in Section 4.2 to Section 4.6, are generic, and the trends
obtained are valid for external galaxies as well.
In this approach the stars, the interstellar atomic hydrogen gas (HI),
and the interstellar molecular gas (H2 ) are treated as three separate disk
components. For simplicity, these are assumed to be concentric, co-spatial
and coplanar w.r.t the galactic mid-plane (z = 0). The steady-state disk
distribution is taken to be symmetric w.r.t. the mid-plane (z = 0). The
disk is embedded within an extended dark matter halo which is taken to
be concentric with the disk. The interstellar gas is taken to consist of two
components: atomic hydrogen gas, HI; and molecular hydrogen gas, H2 .
This is necessary as these are known to have different velocity dispersions
and surface densities; also, the values of these parameters as well as the
fraction of gas in HI or H2 form varies with radius (see Section 2). Hence
for a realistic representation of the Galactic disk, it is essential to formulate
it as a three-component disk. The self-gravity of each disk component is
taken into account. Thus, the Galactic disk is treated as a gravitationally
coupled, three-component system (consisting of stars, interstellar HI and H2
gas), which is in the gravitational field of the dark matter halo.
The disk is assumed to be axisymmetric. Each disk component is as-
sumed to be in a hydrostatic equilibrium along the vertical direction. That
is, for each disk component, its internal pressure gradient balances the net
gravitational force due to all the disk components and the halo, along the z
direction. Each disk component is taken to be isothermal and is character-
ized by a different velocity dispersion. Due to its lower dispersion, gas forms
a thinner layer along the z direction and thus the gas layers are embedded
within the stellar disk of a higher thickness. Further, the molecular gas, H2 is
embedded within the HI gas. See Fig. 5 for a schematic diagram of the dis-
tribution of the three disk components: namely, stars, HI and H2 , embedded
within the dark matter halo.
The self-consistent vertical density distribution for stars and gas in the

36
 Dark Matter Halo

Galactic Disk

Stars

HI
H2
z=0

Figure 5: Schematic diagram of the distribution of the three disk components: namely,
stars, interstellar atomic hydrogen gas (HI) and molecular hydrogen gas (H2 ), in a typical
galactic disk. The radial extent of the disk is much larger than its vertical thickness (not
drawn to scale). The three disk components are taken to be axisymmetric, concentric,
coplanar, and symmetrically distributed w.r.t. the disk mid-plane (z = 0). The stellar
scale height is higher than the HI scale height, which in turn is higher than the H2 scale
height, as observed. These relative values are set by their observed velocity dispersion
values. The disk is taken to be embedded within an extended, massive dark matter
halo which is concentric with the disk; the size and shape of the halo are not yet well-
understood. (The relative sizes of the disk and halo shown are not to scale. The shape of
halo is assumed to be oblate for the sake of illustration).

37
joint potential of the multi-component disk and the halo is obtained. The
additional gravitational force due to the other disk components and the halo
in the coupled case is shown to modify the vertical density distribution of
each disk component, constraining it to be closer to the disk mid-plane. Since
the interstellar gas forms a thinner layer than the stars; therefore, despite
its low mass fraction in the disk, it is shown to exert a strong gravitational
force on the stars near the mid-plane. This significantly affects the vertical
density distribution of stars, as discussed later in this section.
A general point to note is that, in this model, the two gas components are
distinguished from each other, and from the stellar component, only in terms
of their respective vertical velocity dispersion. That is, this model does not
take account of the dissipational nature of the gas components.

4.2. Multi-component disk plus halo model: Thin disk

4.2.1. Formulation of equations: Thin disk
The equations describing the vertical structure for the thin disk case,
under the general assumptions (described above in the basic outline of the
model), were formulated by [16], and are given next. The steady-state, self-
consistent vertical density distribution, ρ(z), for each disk component is ob-
tained by solving together the equations for the hydrostatic balance for each
disk component and the joint Poisson equation for the disk-halo system for
the thin disk case.
It is necessary to take account of the effect of the dark matter halo in the
study of the disk vertical structure. The inclusion of the halo complicates the
formulation of the problem but also makes it richer. We note an important
general point that for two different gravitating structural components such
as the disk and the halo, the density, ρ, and the potential, Φ, for each satisfies
its own Poisson equation locally and a linear superposition describes the net
system. This will be used in writing the joint Poisson equation (Eq. (4.2))
later.
The force equation or the Euler equation or the equation of hydrostatic
equilibrium along the z-axis for each disk component is given as follows,
generalizing the one-component case (for the latter, see [87]; or, Section 3.1):

⟨(vz2 )i ⟩ ∂ρi
= (Kz )s + (Kz )HI + (Kz )H2 + (Kz )h (4.1)
ρi ∂z
1/2
where ρ is the mass density, and ⟨(vz2 )⟩ = σz is the velocity dispersion of
a disk component along z at a given radius R, and (Kz ) = −∂Φ/∂z is the

38
force per unit mass along the z-axis, where Φ is the corresponding potential.
The subscript i = s, HI, and H2 denotes these quantities for stars, HI and
H2 respectively. The last term on the right hand side denotes the force per
unit mass along the z direction due to the dark matter halo. The right hand
side of Eq.(4.1) denotes the total vertical force due to all the components.
This can modify or redistribute the density (ρ) and the potential (Φ) of each
particular disk component.
Thus each disk component feels the gravitational force due to all the disk
components and the halo. This includes gas being affected by stars, even
though the stellar disk thickness is higher than that of the gas. This is
possible because of the open geometry of the disk distribution (see Appendix
A in [121]).
Since the disk is thin and much less massive compared to the halo, its
effect on the vertical distribution within the halo is neglected, so the halo
is taken to remain undisturbed by the disk. Thus, the halo is treated as a
reservoir that just provides an external gravitational field acting on the disk.
The joint Poisson equation for the above system is considered next. For a
thin, axisymmetric disk, only the z component of the Laplacian for the disk
components (stars and gas) needs to be retained in the Poisson equation for
the disk (see Eq. (2.57) from [2]; also, see Section 3.1).
The joint Poisson equation for the multi-component disk plus halo in the
thin disk case is then given by:

∂ 2 Φs ∂ 2 ΦHI ∂ 2 ΦH2 ∂ 2 Φh
 
1 ∂ ∂Φh
+ + + R +
∂z 2 ∂z 2 ∂z 2 R ∂R ∂R ∂z 2 (4.2)
= 4πG (ρs + ρHI + ρH2 + ρh )

Since the halo is assumed to be not affected by the disk, the above equa-
tion reduces to:

∂ 2 Φs ∂ 2 ΦHI ∂ 2 ΦH 2
+ + = 4πG (ρs + ρHI + ρH2 ) (4.3)
∂z 2 ∂z 2 ∂z 2
Thus, in a thin disk, the disk Poisson equation and the halo Poisson equa-
tion are effectively decoupled. However, the disk and halo are still coupled
gravitationally, because of the coupling between a disk component with the
other disk components and the halo through the equation of hydrostatic bal-
ance (Eq. (4.1)). This leads to a redistribution of the vertical density profile

39
for each disk component. In the above equations (Eq. (4.1), and Eq. (4.3)),
ρi (z) denotes the re-distributed or modified density for each disk component
in the multi-component disk plus halo case.
From Eq. (4.3) it can be seen that the force and the resulting den-
sity distribution are functions of z alone, analogous to the one-component
case considered in Section 3.1. Hence the value of ρi (z) depends only on
the local conditions at the given radius. Thus, the problem reduces to a
one-dimensional, and hence a local, problem of determining the density dis-
tribution, ρi (z), at a given radius. Hence, the vertical density distribution
of a given disk component, is only affected by the local gravitational force
of itself, the other disk components, and the halo at a given R (which only
depend on the local values of their parameters). Since the molecular gas is
mainly confined to the inner disk, it affects the stellar vertical density dis-
tribution in the inner disk; whereas the effect of HI on the vertical stellar
distribution is seen in the outer disk as well (see Section 4.2.2 to Section
4.2.6).
Combining Eq.(4.1) and Eq.(4.3) gives the following joint Poisson-hydrostatic
balance equation for the coupled case, which governs the self-consistent ver-
tical density distribution, ρ(z), of a disk component at a given radius:

2
∂ 2 ρi
  
ρi ∂(Kz )h 1 ∂ρi
2
= −4πG (ρs + ρHI + ρH2 ) + + (4.4)
∂z ⟨(vz2 )i ⟩ ∂z ρi ∂z
where the square brackets contain the terms that arise due to the joint po-
tential of the three disk components and the halo in the coupled system.
Even though this is the same for all the disk components; its effect on the
resulting density distribution of each disk component is different due to the
different vertical velocity dispersion for each disk component, as shown later
in this section.
General comments
1. It is easy to see that when there is only one gas component present,
we can set Σ = 0 for the other gas component and then the above equation
reduces to that for a two-component (stars plus gas) disk.
2. [16] treated stars to be a single component for simplicity, as is routinely
done in the study of stellar disk structure in the literature (e.g., [22]). In the
literature, the sech2 result (Eq.(3.4)) is routinely applied to a one-component
stellar disk (e.g., [22]), so that all the stars of different types are treated in
a cumulative fashion. This is a reasonable assumption since most of the

40
disk mass is in stars of type G-K-M [18]. It would be worth treating a
multi-component stellar disk if the values of the surface density and velocity
dispersion for each such stellar sub-component are known observationally. In
that case, it would be straightforward to extend the above analysis to include
a multi-component stellar disk (see Section 8 where this is listed as a possible
future problem).

4.2.2. Numerical solutions

The three coupled second-order differential equations (represented by Eq.
(4.4) are solved simultaneously to obtain the self-consistent vertical density
distribution of each disk component. Each second order differential equation
is split into two first order differential equations for the sake of simplicity.
These are solved numerically as an initial value problem, using the fourth-
order Runge-Kutta method of integration [117]. For each disk component,
the boundary conditions applied at the mid-plane, z = 0 are:
dρi
ρi = (ρ0d )i ; and =0 (4.5)
dz
For a realistic distribution close to the mid-plane, the density distribution
along the z axis is homogeneous close to the mid-plane, hence dρi /dz = 0.
Now, (ρ0d )i , the mid-plane density of the re-distributed matter for a given
disk component i is not known a priori. However, the observed surface density
Σi (R) at a given R is known, and can be used to constrain the value of (ρ0d )i
as explained next. The surface density is twice the area under the curve ρi (z)
vs. z. Given the observed value of Σi (R), the value of the mid-plane density,
(ρ0d )i can be found by trial and error. Once this is obtained, the solution for
the vertical density distribution, ρi (z), is determined easily. In retrospect, a
similar approach was taken to determine the mid-plane density, and hence
the solution for ρ(z), for one-component case ([87]; also, see Section 3.1).
As seen above (in Section 4.2.1), here the distribution of matter can be
treated as a one-dimensional, local problem along the z-axis. Hence the
surface density, Σi (R) for each disk component in the coupled case will re-
main unchanged even when the joint gravitational potential is considered.
This is an important point and this occurs because the gravitational cou-
pling between the disk components causes a redistribution of matter within
a component only along z, but not along R. Hence, even in the coupled case,
one can still use the local Σ(R) as a constraint to obtain the corresponding
re-distributed mid-plane density value, (ρ0d )i at a given point.

41
 All the three disk components: namely, stars, HI and H2 ; affect the den-
sity distribution of each other via the coupled equations (Eq. (4.4)). At each
R, the solutions for the three density functions are obtained simultaneously
by taking account of the effect of the other components in an iterative fash-
ion. See [16] for the details of this iteration procedure, where the effect of
the three components is added successively. The above iteration cycle is re-
peated four times until each of the distribution converges with a fifth decimal
accuracy. An important feature of this scheme is that the above procedure
yields the solution for the self-consistent vertical density distribution for all
three disk components (stars, HI and H2 ) simultaneously.
The density distribution of each component will be affected by the joint
gravitational potential, and would no longer be give by a sech2 function
as for a one-component case. [16] point out that they obtain a modified,
sech2 -like distribution for each component. They define the HWHM (half-
width-at-half-maximum) of the distribution to be the vertical scale height.
This definition was also used in the earlier paper on the disk distribution
affected by a molecular cloud complex [112]. The HWHM of the resulting
distribution is refereed to as the vertical scale height or disk thickness in the
subsequent papers that studied this model. This was a pragmatic choice, and
gives the actual value of the thickness measured – that is not related to any
presupposed or ad hoc functional form. See Appendix A for further detailed
discussion of HWHM, and how it is a robust indicator of disk thickness.
The above numerical calculation to obtain ρi (z) was repeated at different
R values, to obtain the modified or redistributed density distribution, ρi (z),
and hence the resulting scale height (HWHM) vs. R for each disk component
in the coupled case.

4.2.3. Application to the Milky Way

The above model was first formulated in general terms and then applied
to the Milky Way in the inner Galaxy R < 12 kpc) by [16]. The scale heights
are shown to be smaller for each disk component compared to the corre-
sponding one-component values; due to the additional gravitational force in
the coupled case. These results are in a good agreement with observations
for all the three disk components studied, as discussed next. This problem
was studied in greater detail using the same model by [19], who explored
the various features of the re-distributed stellar density distribution in the
coupled case. Further, they covered a larger radial range (4-22 kpc), with a

42
particular focus on the stellar distribution in the outer disk. It was found
that the gas plays the dominant role in constraining the disk vertically in
the inner Galaxy; while the halo has the dominant constraining effect in the
outer Galaxy (beyond R=18 kpc), as will be discussed next. The main results
from these studies are summarized below.
Parameters used
The input parameters needed to solve the coupled equations (Eq. (4.4))
are the surface density and velocity dispersion as a function of radius for
the stars, HI and H2 gas. These values are taken from observations and are
briefly given here, see [16] for details of the observed values. The mass model
for the Galaxy by [82] was used as being modern and consistent with other
observations at that time, such as the local surface density as given by [94].
On subtracting the total observed gas density [70]; the local stellar surface
density at the solar neighbourhood, R= 8.5 kpc, is obtained as 45 M⊙ pc−2 .
An exponential stellar disk distribution, with a radial disk scale length, RD
= 3.2 kpc is adopted from [82]. This gives the central stellar surface density
as 640.9 M⊙ pc−2 .
The observed stellar radial velocity dispersion values from R= 1 to 17
kpc in the Galaxy were measured by [56] (see Section 2.1.4). These fall
exponentially with radius with a scale length, Rv = 8.7 kpc. Assuming
the same ratio of the vertical to radial velocity dispersion as in the solar
neighbourhood (= 0.45), gives the vertical stellar velocity dispersion values
as a function of radius. Thus, the vertical velocity dispersion is also taken to
fall exponentially with radius with the same scale length, Rv = 8.7 kpc. [19]
assume the same rate of fall-off with radius in the outer Galaxy, until the
calculated stellar velocity dispersion is ∼ close to the gas velocity dispersion.
Beyond this radius, the stellar velocity dispersion is taken to saturate at this
value. This physically motivated assumption is based on the fact that the
stellar velocity dispersion cannot be smaller than the dispersion in the gas
from which stars form. This occurs at R=17 kpc in the Milky Way, for the
observed HI gas dispersion of ∼ 7 km s−1 , as discussed next. This point will
be discussed more in Section 5, along with the observational evidence for
the saturation in the stellar velocity dispersion at large radii [122] – which
confirms the above assumption by [19].
The surface density values for HI and H2 are taken from [70]. In the work
reported here by [16] and [19], the vertical HI gas velocity dispersion is taken
to be 8 km s−1 in the inner disk (up to R=12 kpc). This is based on the

43
value given by [31], and as measured for ∼ 200 face-on galaxies by [123]. The
HI gas dispersion is observed to slowly fall and saturate to 7 km s−1 at large
radii [124, 125]. Hence, the HI dispersion is assumed to fall gradually beyond
R=12 kpc at the rate of - 0.2 km s−1 kpc−1 . This tapers off to a value slightly
higher than 7 km s−1 at R=17 kpc; and is assumed to be constant beyond
this radius, see [19]. The model scale height results are compared with the
observed values. The observed scale height values for HI for R < 8.5 kpc are
taken from [108]; and those beyond R > 8.5 kpc (corrected for the Galactic
warp) are taken from [42].
The dark matter halo is taken to have a pseudo-isothermal density profile:
ρ0h
ρ(R, z) = (4.6)
(1 + (R2+ z2 )/Rc 2 )
following the mass model of [82] that was adopted. Here ρ0h and Rc are the
halo central density and the core radius respectively, whose values are ρ0h =
0.35 M⊙ pc−3 , and Rc = 5 kpc.

4.2.4. Results for vertical scale heights in the inner Galaxy

The resulting vertical scale heights for the three disk components in the
inner disk (R < 12 kpc) and a comparison with observations (from [16]) are
briefly discussed next. This radial range clearly illustrates the basic features
of a multi-component disk model. Although the dark matter halo was in-
cluded in this work in order to be comprehensive, it does not contribute much
to the determination of the stellar distribution in the radial range considered.
For example, even at R=12 kpc, the largest value of the radius which was
considered in that work; the inclusion of the halo affects the HI scale height
(HWHM) by only 12%. The effect of the halo on stellar distribution becomes
important, even dominant, at large radii [19], as discussed later in Section
4.2.5 and Section 4.2.6.
The resulting value of the HWHM for each disk component, responding
to the joint potential in the coupled case, is plotted (as a solid line) in Fig. 6,
while the dashed line corresponds to the HWHM values when each component
responds to the stellar potential alone (i.e., with no self-gravity in the case of
HI and H2 ). The three panels in Fig. 6 contain results for HI, H2 and stars,
respectively.
The most striking result is that for each disk component (stars, HI and
H2 ), the joint potential approach gives a smaller scale-height at each radius
than the value obtained using stellar potential alone. This effect is stronger

44
Figure 6: The plot of scale height (HWHM) vs. R for interstellar HI, H2 and stars vs. R,
in the three panels respectively, at R=8.5 kpc. In each case, the solid curve denotes the
theoretical result obtained under the joint potential; and the dashed curve represents the
theoretical result obtained under the stellar potential alone, i.e., with no self-gravity in case
of HI and H2 . For each disk component, the scale height values under the joint potential are
significantly lower, compared to under the stellar potential alone - particularly at large
radii; and show an overall better agreement with observations. In particular, the joint
potential case explains the old puzzle [109] of nearly constant HI scale height observed in
the inner Galaxy. See Fig. 7 for an improved fit to the HI data. Source: Taken from [16]

45
at larger radii. Thus, including the gas self-gravity, and the additional grav-
itational force due to gravitational coupling between the disk components,
decreases the resulting scale heights and brings them closer to the observed
values. The detailed comparison is given next.
Consider the HI case shown in Fig. 6 first. The dashed line representing
the HWHM as a response to the stellar potential alone, increases exponen-
tially with radius; and thus deviates strongly from the observed values beyond
8 kpc. On using the joint potential, the scale height reduces significantly, es-
pecially at large radii; because the HI gas gravity becomes important. Hence,
the model scale heights then show an overall agreement with observations,
especially in the middle range of 5-10 kpc. Next, a small linear gradient of
-0.8 km s−1 kpc−1 was tried, starting with a value of HI velocity dispersion
of 8 km s−1 at R = 8.5 kpc. The resulting scale height values vs. R for HI,
H2 and stars are given in the three panels in Fig 7. This value of the velocity
gradient was chosen since it was found to give the best-fit between the model
HWHM and the observed values for HI for the range 2-12 kpc (see the panel
for HI in Fig. 7). On comparing the panel for HI in Fig 7 with panel for HI
in Fig. 6, it is clear that including the HI velocity gradient gives a better
overall agreement with the observed data. A plausible explanation for this
radially varying HI velocity dispersion (with a higher velocity dispersion at
smaller radii) is the higher energy input due to the supernova input [74].
This rate is expected to be higher in the inner region, where the molecular
gas – which forms the site of star formation – is seen to dominate. Thus this
model explains the old puzzle of the nearly constant HI scale height in the
inner Galaxy [109].
For H2 as well, neglecting the gas gravity (dashed line) gives a large
deviation from the observed values, see panel for H2 of Fig. 6. On including
the gas gravity, and using the joint potential in the gravitationally coupled
case, the theoretical results for HWHM (solid curve) agree very well with the
observed values, see [70]. Thus, this model gives a physical explanation for
the vertical scale height distribution of the scale heights of the H2 gas in the
Galaxy, which had not been studied before in the literature.
For stars, the stars-alone potential gives an exponentially increasing curve
for HWHM versus R (solid line) (see the panel for stars, Fig. 6). In contrast,
the stellar scale height curve in the joint case shows lower values, with a
nearly flat behaviour (with a constant value of HWHM of ∼ 300 pc) up to
R = 5 kpc. Between R= 5-10 kpc, it shows a moderate increase with the
best-fit gradient of 24 pc kpc−1 . Beyond 10 kpc, the curve corresponding to

46
Figure 7: The plot of scale height (HWHM) vs. R for HI, H2 and stars vs. R in the three
panels as shown, at R=8.5 kpc; where a radial gradient in HI velocity dispersion with a
value of -0.8 km s−1 kpc−1 has been included in the calculations. The resulting HI scale
heights are in good agreement with observations over the entire radial range studied, and
the fit is better than in panel for HI of Fig. 6 . Interestingly, the values of scale heights
for H2 and stars remain nearly unchanged on introducing the gradient in the HI velocity
dispersion (compare Fig. 6, panels for H2 and stars with Fig. 7, panels for H2 and stars,
respectively) – see the text for the physical explanation. Source: Taken from [16]

47
the joint potential saturates at 480 pc. Thus, the additional gravitational
force due to gas and the dark matter halo in the joint system decreases the
amount of flaring in the stellar disk to a moderate value. For example, at
R= 8.5 kpc, the stellar potential alone gives a stellar scale height of 550 pc,
but the inclusion of gas gravity reduces it to 380 pc. This reduced value is in
a good agreement with directly observationally determined value of 350 pc
for the Milky Way ([2]; also, see Appendix A).
The variation of stellar scale height with radius cannot be compared with
optical observations in our Galaxy due to dust extinction. However, the
observed near-IR data available at that time, namely from Spacelab2 also
showed a moderate increase in scale height with radius with the best-fit value
of 20 pc kpc−1 in this radial range [126]. This trend is in a good agreement
with the model results for the coupled case, as given above.
It is interesting that the stellar disk is not strictly flat as has claimed by
[22]; rather, it shows a small but finite increase in scale height within the
optical disk. Such a moderate flaring of the stellar disk is a generic result,
and will be discussed further in Section 5.
This approach, studied using realistic input parameters, cohesively and
naturally explains the observed scale height distributions of all the three disk
components: namely, stars, HI and H2 , in the inner region (R=2-12 kpc).
This was the success of this model.
An interesting result to note is that when the velocity gradient in HI ve-
locity dispersion as given above is used, the corresponding resulting H2 and
stellar scale height distributions respectively in the coupled case do not show
any noticeable difference (compare panels for H2 and stars in Fig. 6 with the
corresponding ones in Fig. 7 respectively). That is, the change in velocity
dispersion of HI gas affects the HI distribution but has little effect on that
of the H2 or stellar disk. This is because the Jeans equation describing the
pressure-equilibrium, or the equation of hydrostatic equilibrium, of a given
component only depends on its own velocity dispersion (Eq.(4.1)). Hence,
the velocity dispersion of each component only directly affects its own scale
height (see Eq.(4.4)). Therefore, a change in the velocity dispersion of HI
only indirectly affects the density distribution of other coupled disk compo-
nents, through the change in the density distribution of HI whose velocity
dispersion has been changed (see Eq.(4.4)). This is an important physical
point and it comes about because the pressure support in the vertical Jeans
equation only depends on the vertical velocity of that particular component,
whereas each disk component feels the joint gravitational force due to all

48
the coupled disk components. This is analogous to the study of stability
of linear, axisymmetric planar perturbations in a two-fluid (stars plus gas
disk), where the support in the radial Euler equation for a component only
depends on the pressure term due to itself, but each component feels the net
gravitational force due to both the disk components [121].

4.2.5. Physical understanding of re-distribution of stellar density

The results in Section 4.2.4 show a decrease in the vertical scale height
(HWHM) for each disk component in a joint potential compared to that in
the stellar potential. The physics behind this is illustrated clearly in the
study by [18]. They considered a three component (stars, HI and H2 gas)
disk where the focus was to show the dynamical effect of the low dispersion
component, namely gas, on the stellar vertical density distribution in the
coupled case.
First, consider each isothermal disk component separately, with its den-
sity distribution, ρ(z), and the corresponding force per unit mass, Kz , as
given respectively by Eq. (5) and Eq. (6) from [18]; or, see Eq.(3.3), and
Eq.(3.5). Fig. 8 gives a plot of |Kz | vs. z shown on a logarithmic scale for
stars, HI and H2 at R=6 kpc. This shows that their magnitudes are com-
parable; and, in fact, up to |z| < 150 pc, the force due to gas is significant,
∼ 30% of that due to stars. This is because of the low velocity dispersion of
gas, which confines gas to a thinner layer; so that it is concentrated closer
to the Galactic mid-plane than the stars. The gas mass fraction at R=6
kpc is 16% of stellar surface density (see Table 1 from [18]). Thus, the self-
gravitational force close to the disk mid-plane is a non-linear function of the
gas surface density in the disk. Hence despite its low mass fraction, the gas
can have a significant effect on the vertical density distribution of the main
mass component, namely, the stars, as shown next.
The stellar density distribution in the coupled case is obtained numeri-
cally as a solution of the coupled joint Poisson-hydrostatic balance equations
(see [18]). Fig. 9 shows the model stellar vertical density distribution for
the stars-alone case and also that in the coupled case. This clearly shows
the dynamical effect of gas gravity on the vertical stellar distribution. In
the coupled case, the central density of the stellar distribution is higher, by
∼ 25%; and the thickness (HWHM) is smaller, by ∼ 20%; compared to the
stars-alone case.
This can be physically understood as follows. The additional gravitational
force due to the other disk components modifies the stellar distribution, so

49
Figure 8: Plot of vertical self-gravitational force per unit mass, |Kz |: for stars-alone; H2 -
alone; and HI-alone cases; drawn on a log scale vs. z, at R=6 kpc. The force due to H2
gas is a significant fraction, ∼ 30%, of the force for the stars-alone case; for z values close
to the Galactic mid-plane (|z| < 150 pc). Source: Taken from [18]

that it is vertically constrained to be closer to the mid-plane. Thus the mid-

plane density is higher compared to the stars-alone case. For a constant
surface density, this therefore results in a smaller thickness (HWHM); and at
the same time, the shape of the density profile is also altered, which becomes
steeper.
Thus, the redistributed stellar distribution in the coupled case has a
higher central density, its thickness is smaller, and the shape of its density
profile is steeper; compared to the stars-alone case. This is the generic be-
haviour of redistribution of any gravitating disk component due to additional
gravitating components (disk components or halo) in the system.
While we have focused so far on the effect of gas on the stellar density
distribution; similarly, the gas density will also be redistributed, due to the
gravitational force of stars and halo, and will be constrained closer to the
mid-plane than for the gas-alone case (see Section 4.2.6, and Appendix A
for further discussion on this).7 Thus even though each isothermal disk

7
The case in [16] (or Section 4.2.4) is somewhat different because first the gas response
to stellar potential is considered where the gas gravity is not included and then the gas

50
Figure 9: Self-consistent vertical density for stars vs. z: for the stars-alone case; and
the stars in the gravitationally coupled case (stars, H2 , and HI), at R=6 kpc. Due to
the additional gravitational force of gas, the stellar distribution in the coupled case is
constrained closer to the mid-plane, such that: its mid-plane density is higher; the scale
height (HWHM) is smaller; and the density profile is steeper – compared to the one-
component, stars-alone case. Source: Taken from [18]

component by itself would satisfy a sech2 distribution; its net vertical density
distribution in the coupled system is different from sech2 , see Section 4.2.6
for further discussion on this.
A detailed quantitative analysis of the redistribution of the stellar vertical
density distribution, which includes the gravitational effect of gas as well as
the dark matter halo; with a particular focus on the outer Galaxy, was carried
out by Sarkar & Jog (2018) [19], as briefly summarized next. The values of
Kz for stars-alone and gas-alone are obtained as done above, or see [19] for
details. For the dark matter halo, Kz is obtained by taking the z derivative
of the halo potential (see Eq. 7, [19]). The values for |Kz | are plotted on a
log scale for stars, H2 , HI and dark matter halo vs. R in Fig. 10 for R = 6
and 18 kpc respectively. The choice of one radius in the inner Galaxy and
one in the outer Galaxy helps bring out the dynamical result that the effect

distribution in a coupled (stars plus gas plus halo) case is considered that also includes
gas gravity - see Fig. 6 panels for H2 and stars.

51
 2 0
10 10

1 (a) (b)
10
−1
10
0 Stars Stars
10
H2 DM

−2
−1 10
10
|Kz|

DM

|Kz|
−2 HI
10 −3
10
HI
−3
10
R = 6 kpc −4 R = 18 kpc
10
−4
10

0 50 100 150 200 250 300 350 400 450 500 0 200 400 600 800 1000 1200
Vertical Distance z (pc) Vertical distance z (pc)

Figure 10: Plot of gravitational force per unit mass, |Kz |, exerted by each disk component
(stars, HI and H2 ) and the dark matter halo (DM) separately, drawn on a log scale vs. z,
at R= 6 kpc (left panel) and R=18 kpc (right panel). This figure shows that in the coupled
case, the stellar distribution will be mainly affected by the H2 gas gravity rather than by
the halo or HI in the inner Galaxy, at R=6 kpc. In contrast, in the outer Galaxy, at R=18
kpc, the force due to the dark matter halo dominates over the stellar self-gravitational
force itself; hence, the halo will strongly affect, and in fact, mainly determine the modified
stellar distribution in the coupled case. This is confirmed by the results shown in Fig. 11.
Source: Taken from [19]

of gas and halo on stars dominates at different radial regions – namely, in

the inner and outer regions, respectively.
Fig. 11 shows the resulting vertical stellar density distribution in the
stars-alone case, and then in the coupled case (the latter obtained by solving
Eq. 1 from [19]) at R= 6 and 18 kpc respectively (see the left and right
panels, respectively in Fig. 11). The plot at R=6 kpc is a generalization of
Fig. 3 of [18], that is, Fig. 9; since in Fig. 11 the effect of halo on the stellar
vertical distribution is also included. In the coupled case, first only the effect
of dark matter halo is included; then the effect of dark matter halo as well
as gas is included. This step-by-step addition of different components allows
a study of the differential effect of gas and dark matter halo on the stellar
vertical distribution.
As seen in Fig. 10, at R=6 kpc, the force |Kz | due to H2 dominates over

52
 0.16 10 -3
2
S+DM+G
S+DM S+DM+G
(a)
(c)
0.12
1.5
Stars R = 6 kpc S+DM
R = 18 kpc

0.08 1

0.04 0.5 Stars

0 0
0 100 200 300 400 500 0 200 400 600 800 1000 1200
Vertical Distance z (pc) Vertical Distance z (pc)

Figure 11: Self-consistent vertical density of stars vs. z at R=6 kpc (in the inner Galaxy)
(left panel), and at R=18 kpc (in the outer Galaxy) (right panel). The three curves rep-
resent the density distribution of stars in the gravitational field of: stars-alone; stars plus
dark matter halo; and stars plus dark matter halo plus gas – denoted respectively by the
dashed-dot, dashed and solid curves, respectively. The inclusion of other gravitating com-
ponents (gas and halo) results in an overall constraining of the stellar distribution closer
to the mid-plane: such that, the mid-plane density is higher; the scale height (HWHM) is
smaller; and the density profile is steeper; compared to the results for the stars-alone case.
The constraining effect on stars is mainly due to gas in the inner Galaxy (left panel), while
the dark matter halo has a dominant effect in the outer Galaxy (right panel). Source:
Taken from [19]

that due to the dark matter halo; hence the main constraining effect or the
redistribution of the stars is due to H2 gas, while the halo has a small effect.
This is confirmed by the left panel in Fig. 11 where the HWHM is reduced
by ∼ 6% due to the effect of the halo as compared to the stars-alone case;
whereas the addition of gas (H2 and HI) reduces the HWHM by further 20%.
Thus, gas has a higher effect on the stellar distribution (see Table 2 from
[19] for the numerical values of HWHM) than the halo. The corresponding
increase in the mid-plane density for stars due to the gravitational effect of
dark matter halo and gas is 7% and 33 % respectively (see Table 2, from
[19]).
At R=18 kpc, on the other hand, the force |Kz | due to dark matter halo
dominates that due to the self-gravitational force of stars and HI; there is no
H2 gas at this radius. Here, clearly the halo would play the dominant role in

53
shaping the modified stellar vertical distribution. This is confirmed by Fig.
11, right panel. Here the gravitational force of halo decreases the HWHM of
the stellar distribution by a huge factor of ∼ 2.3 compared to the stars-alone
case; while including gas decreases it further by ∼ 20% only (see Table 2 in
[19]). In fact, beyond R=17 kpc, the constraining effect of halo dominates,
and the halo is the main determinant of the stellar vertical distribution (see
Table 2 from [19])). In the outer disk, the stellar vertical density distribution
is steeper than the stars-alone case mainly due to the effect of the halo.
Thus, gas has an important constraining effect on the redistribution of
stellar vertical distribution at small radii; while the halo has the dominant
constraining effect in the outer Galaxy.

4.2.6. Detailed results for stellar density distribution in the outer Galaxy
The quantitative results for the three features of the redistributed stellar
density distribution (arising due to the constraining effect of gas and the
halo): namely, the mid-plane density, the scale heights (HWHM), and the
vertical stellar profile, are given next. The focus is on the outer Galaxy. The
dark matter halo plays a significant role in this radial range, confirming what
was discussed above.
Measuring the constraining effect: Disk thickness
A plot of the HWHM of the vertical stellar distribution, for the multi-
component disk plus halo system vs. radius, for R=4-22 kpc is given in Fig.
12; also, see Table 2 in [19] for the numerical values. Fig. 12 shows that in
the coupled case, the stellar scale height increases by a moderate amount, by
a factor of ∼ 50% from R=4 to 16 kpc. Beyond R = 17 kpc, the stellar disk
flares steeply because the stellar velocity dispersion saturates (by choice) to
the gas dispersion value beyond this radius (see Section 4.2.3). As can be
seen from Table 2, [19], the stellar disk by itself would flare by a factor of
13.6 from R= 4 to 22 kpc. The inclusion of gas, and mainly the halo (which
is effective at larger radii), reduces the flaring to a more moderate value of
a factor of 3.3. The gas and mainly the halo restrict the thickness to be
< 1 kpc even at R = 22 kpc. The lower disk thickness would make the disk
robust, that is, help resist distortion due to tidal perturbations ([19]; also,
see Section 4.6).
We point out that the various observations of the Milky Way outer stellar
disk do show flaring with radius: [50] using 2MASS data for red clump and
red giant star; [41] using SDSS-SEGUE data for F8V-G5V type stars, and

54
 900

800 HWHM

HWHM (pc)
700

600

500

400

300

200
3 5 7 9 11 13 15 17 19 21 23
Radius (kpc)

Figure 12: Plot of the model scale height (HWHM) of the vertical stellar density distribu-
tion in the joint potential of the disk and dark matter halo vs. R in the Galactic disk. The
stellar disk thickness increases gradually until about R=17 kpc and then flares beyond
that in the outer disk. Source: Taken from [19]

[52] using LAMOST data on red branch stars. These agree reasonably well
with the results in Fig. 12 (also, Table 1, column 4 in [19]). The results of
[41] show somewhat higher flaring. The higher values in [41] could be partly
due to a possible contamination by thick disk stars ([127]; also, see Section
5.3). It is worth noting that a similar trend of flaring of stellar disks in the
outer Galaxy was noted and discussed by [128], who had also questioned the
constancy of the stellar scale height with radius. The detailed physics and
implications of the flaring stellar disk will be discussed later in Section 5.
[19] also propose a new parameter: z1/2 , the half-mass scale height as
measured from the mid-plane as another way of measuring the disk thick-
ness (also, see Appendix A). This can be used as indicator of the constraining
effect of gas and halo. The values of z1/2 are comparable to but somewhat
smaller than the HWHM values (see Tables 2 and 3 in [19]). From an obser-
vational perspective, z1/2 can be thought of as an indicator of the half-light
scale height for a constant M/L (mass-to-light) ratio, for an edge-on galaxy.
A similar parameter has also been proposed and used by [129] to analyze the
data from TNG simulations.

55
Measuring the constraining effect: Mid-plane density
The mid-plane density of each disk component in a coupled system is
higher compared to its one-component value, due to the constraining effect
of the other disk components and the halo (see Section 4.2.5). The mid-plane
stellar density in the multi-component disk plus halo model is higher by 50%
at R=8.5 kpc; while the increase is much higher, by a factor of ∼ 3 − 4 at
R=18, 20 kpc respectively (see Table 4, [19]). This is because at large radii,
|Kz |, the force due to the halo dominates that due to the stellar self-gravity
itself. While the mid-plane stellar density is not likely to be directly observed,
its higher value can have interesting dynamical consequences. For example,
the vertical oscillation frequency normal to the plane, as given by (4πGρ0 )1/2
(e.g., [2]), would be higher by a factor of ∼ 2 at R=18-20 kpc. This could lead
to a better-mixed vertical distribution in the outer Galactic disk. A similar
increase in mid-plane gas density (details are given later in this section) would
help increase star formation seen in the outer Galaxy, which is otherwise hard
to explain. Thus, the dark matter halo has a dominant influence on the disk
vertical structure in the outer disk.
Measuring the constraining effect: Stellar vertical profile
In real galaxies, the observed distribution near the mid-plane shows an
excess over the sech2 distribution, and the distribution typically obeys a
sech or an exponential distribution (Section 4.1). Hence [14] suggested the
following family of curves that could fit the observational trends:

ρ(z) = 2−2/n ρe sech2/n (nz/2ze ) (4.7)

where n= 1,2 and n→ ∞ correspond to a density profile with a sech2 , sech
and an exponential z distribution respectively. The parameters n, ze , ρe are
obtained by fitting the above function to the data and doing the best-fit
analysis. This function is not based on physical principles, rather it is ad hoc
as noted by [14] – who proposed it (also, Section 4.1.1). Despite this, it has
been used in many observational studies to fit the data on external galaxies
so as to get the best-fit value of n (e.g., [15, 29]), where n is taken to be an
indicator of steepness of the density profile.
The additional gravitational force due to the other disk components and
dark matter halo results in the stellar profile being steeper than the one-
component case (see Section 4.2.5; and, [19]). Note that even though the
stellar disk is taken to be isothermal, the net stellar distribution is no longer
given by a sech2 profile in a coupled, multi-component disk under the gravi-

56
 1.1

1 exponent 2/n

0.9

Exponent 2/n
0.8

0.7

0.6

0.5
R = 6 kpc
0.4

0.3

0.2
50 100 150 200 250 300 350 400 450 500

Fitting range ∆z (pc)

Figure 13: Plot of the best-fit exponent 2/n vs. |∆z| for the range of z values over which
the model vertical density distribution for stars is fitted by a distribution of type sech2/n ;
shown for R = 6 kpc. The value of n varies with |∆z|, hence n is not a robust indicator
of the stellar density profile. Source: Taken from [19]

tational field of the halo. To quantify this effect, [19] tried to fit the model
results to Eq. (4.7). They found that the above expression is not adequate
to explain the results for the model stellar density profiles, as discussed next.
In the above expression, 2−2/n ρe is the mid-plane stellar density which is not
known from observations, but is known from the model calculations. [19] fit
the resulting model density distribution to the above function (Eq.(4.7)) to
get the best-fit values of n and ze . They find that the best-fit value of 2/n
is not robust; instead, it varies with ∆z, the range of z values chosen for
the fitting – see Fig. 13 which shows the results for R= 6 kpc. Thus, the
expression for the density profile, (Eq.(4.7)), which was proposed by [14] is
not physically valid for a realistic galactic disk.
Keeping in mind this caveat about n not being robust; the model density
distribution results are next fit for |∆z| < 150 pc with the above function
(Eq.(4.7)) at different R values; so as to get an idea of the trend, and to
compare with observations where the data close to the mid-plane are studied.
The resulting plot of 2/n vs. R is given in Fig. 14. The result gives n > 1
up to R < 14 kpc. Interestingly, this is the radial range over which the gas is
mainly responsible for constraining the stellar distribution (see Table 2 from

57
 5.5
5
exponent 2/n
4.5
4

Exponent 2/n
3.5
3
2.5
2
1.5
1
0.5
0
2 4 6 8 10 12 14 16 18 20 22 24

Radius (kpc)

Figure 14: Plot of the best-fit value of the exponent 2/n when the model density distri-
bution is fit by a distribution of type sech2/n over a range of z values, |∆z| < 150 pc vs.
R. For radii > 14 kpc, a new range of n < 1 is obtained. This corresponds to the radial
region where the main constraining effect is due to the dark matter halo. Source: Taken
from [19]

[19]). On the other hand, it is found that n < 1 in the outer Galactic disk;
where the constraining is mainly due to the halo. Observations of external
galaxies typically cover inner regions; which explains why the observed values
give the best-fit n to be > 1, corresponding to the density profile between
sech and an exponential. The parameter region n < 1 obtained here is new,
and has not been studied before in the literature. The important point to
stress is that, in both cases, whether n > 1 or n < 1, the density distribution
is steeper than the distribution for the corresponding one-component cases.
It is often stated in the literature that the index n > 1 or (2/n) < 2
can be taken to be an indicator of the steepness of the profile, with higher n
corresponding to a steeper profile [29, 18]. This is misleading since the case
when n < 1 could also denote a steeper profile compared to the particular
one-component case. As shown above (Section 4.2.5), for a constant surface
density, any additional gravitational component (gas or halo) will constrain
the stellar distribution closer to the mid-plane and cause a steepening of the
stellar profile compared to the corresponding one-component stellar disk of
constant surface density, but n could be > 1 or < 1. See for example, the

58
two panels in Fig. 11 where n > 1 for R = 6 kpc, and n < 1 for R= 18 kpc
(as obtained from Fig. 14).
In reality, all the three parameters, namely, the mid-plane density, width
and n, together decide how sharply the stellar density profile falls with z
[91]. Similarly, the other disk components, HI and H2 , will also show a
steeper profile than the corresponding one-component cases (as mentioned
in Section 4.2.5; also, see Appendix A for details).
We have given a detailed discussion of this topic, including the limitations
of using the parameter n as an indicator of steepness of density profile – so
as to give a complete picture; and also because Eq.(4.7) has been a popular
form in the literature to fit the observed excess of intensity close to the mid-
plane. As shown above, n alone is not an indicator of steepness. Moreover, it
has been shown that the value of the best-fit n itself is not robust; rather, it
varies with ∆z, the range used for fitting. Further, n could be < or > 1 and
yet denote a steeper profile compared to the one-component case. Therefore,
the density profile as proposed by [14], see Eq.(4.7), is not physical and not
applicable to a real disk. Hence, it can only be used with some caution to fit
the data.
We highlight an important point that the n value obtained from fitting
the model results is not equal to 1, neither is it very large – which would
correspond to a sech2 or an exponential profile respectively (see Eq. (4.7);
and Fig. 13 and Fig. 14 which show the model results for n). Thus, the
actual model density profile is more complex than given by either of these
limits.
Recall that, for a typical galactic disk, the determination of ρ(z) is a local
problem; and can be used to uniquely trace the corresponding gravitational
potential (Section 3.1, Section 4.2.1, and Section 4.3.1). This can be done
numerically. However, an analytical form to represent the numerical results
for the disk density ρ(z) obtained from the multi-component disk plus halo
model for the various cases; or, the corresponding gravitational potential at
a given R; is not easy to obtain and has not been attempted so far.
Effect on HI gas
The model results for the vertical density distribution of HI gas in the
outer Galaxy (from [19]) are given next. The HI vertical density distribution
for the three cases: HI gas-alone, gas plus dark matter halo, and the coupled
stars plus gas in the halo potential, is given for R= 18 kpc and 22 kpc in
Fig. 15. At there radii, there is no observed H2 gas, so the gas component

59
 10 -3 10 -4
1.75 5

1.5 HI+DM+S HI+DM+S

(a) 4
(b)
1.25 HI+DM

R = 18 kpc HI+DM
3 R = 22 kpc
1

0.75 2

0.5 HI HI
1
0.25

0 0
0 200 400 600 800 1000 1200 0 500 1000 1500
Vertical Distance z (pc) Vertical Distance z (pc)

Figure 15: Self-consistent vertical density distribution of HI gas vs. z in the outer Galaxy,
at R=18 kpc (a), and R=22 kpc (b); under its own self-gravity, then also including the
halo gravitational force, and then also including the gravitational force of the dark matter
halo and stars (shown by dashed-dot, dashed and solid curves respectively). Under its
own self-gravity the HI distribution is extremely extended due to its low self-gravity in the
outer Galaxy. The inclusion of halo gravitational force strongly constrains the HI vertical
distribution in the outer Galaxy, decreasing the thickness (HWHM) by a factor of 3-4, and
the effect is progressively higher at larger radii. The reduced HI thickness values for HI
in the outer Galaxy agree well with the observed values, see the text for details. Source:
Taken from [19]

consists of only HI gas. It is striking that the HI gas-alone under its own
gravity is highly extended vertically; with a HWHM of 1.8 kpc and 3.8 kpc
at R= 18 and 22 kpc, respectively.8
These values are much higher than the observed values [42]. Including the
effect of the gravitational field of the halo substantially reduces the HWHM,
by a factor of 2.7 and 4.3 at these two radii, respectively; and the HI distri-
bution becomes steeper in the coupled case [19]. The decrease is only slightly

8
Note that this treatment of HI gas is different from that in Fig. 6 panel (b) and (c),
where first the gas response to stellar potential is considered without including the gas
gravity; and then the gas distribution in a coupled potential (of stars, gas and halo) that
includes gas gravity is considered.

60
larger on including the effect of stars, so that the net HWHM values are equal
to 502 pc and 798 pc, respectively.
These net resulting HWHM values of HI gas showing flaring (studied up
to R= 22 kpc) agree fairly well with the observed values [42, 110], while [111]
get higher observed values. An important point to note is that without the
confining effect of the halo, the gas distribution would be very extended; and
it would then be more susceptible to disturbance by external perturbations,
such as due to tidal encounters, as well as gas dynamical processes. Thus
the halo cushions the outer disk against getting tidally distorted.
An interesting point is that in the outer Galaxy, beyond R=18 kpc, the
surface density of stars and gas is comparable (see Table 1 in [19]). Also,
their velocity dispersion values are identical, by choice (see Section 4.2.3).
Hence their density distributions are nearly identical. By R= 22 kpc, the
gas surface density dominates (see Table 1 in [19]). In any case, beyond
R=18 kpc, the main gravitational force is due to the halo – which mainly
determines the vertical density distribution of a disk component.
The corresponding HI mid-plane density at R= 18 kpc and 22 kpc in-
creases by a factor of 3.6 and 4.1 respectively, compared to the gas-alone
values. The resulting corresponding mid-plane HI density values are 1.6
×10−3 and 4.1 ×10−4 M⊙ pc−3 , respectively. The gas density is used as an
indicator of onset of star formation. The increased mid-plane density due to
the constraining effect of halo, and the gas being dissipational, makes the gas
more susceptible to star formation (also, see [130]). This effect is higher at
larger radii where the halo is dominant, hence it could explain the star for-
mation seen in outer disks of some galaxies. This is an important implication
from this model, and needs to be explored further.
The gas distribution is constrained towards the mid-plane, and has a
steeper density profile in the coupled case (see Fig. 15, and the discussion
in Section 4.2.5).9 A quantitative measurement of this for gas distribution
needs to be done, in analogy with that done for stars. It is claimed that the
gas distribution in the coupled case is a Gaussian ([131], see Fig. 7 from that
paper). However, an actual detailed fitting to the results has not been done.
See Appendix A for further discussion on this.

9
The physics behind the reduction in HI scale height and the change in the gas profile
in the coupled case was predicted earlier in Section 4.2.5, where an analogous case of
effect of gas on stars in a coupled case was considered which included gravity due to all
the components as done here.

61
 To summarise the discussion so far in Section 4, we stress that the model
developed (Sections 4.1, 4.2.1, 4.2.2) is general, although it is applied here to
the Milky Way, since the parameters are the best-known for it. The results
obtained in Section 4.2.3 to Section 4.2.6 for a thin disk case are generic
and the trends obtained are valid for external galaxies as well. Indeed, the
model will be applied to other galaxies in Section 4.2.7 for the thin disk case.
The results from the thin disk case are applied to obtain the self-gravitational
energy of a multi-component disk in Section 4.6. The model will be developed
for other physical cases and also applied to other galaxies in the following
sections (see Sections 4.3, 4.4, 4.5).

4.2.7. Application to NGC 891, NGC 4565: Radial variation in stellar scale
height
In studies of vertical luminosity distribution of edge-on galactic disks, [22]
measured the vertical scale height of stars and claimed that it is independent
of the radius, R, that is, the radial distance from the centre (Section 2).
They proposed that this implies a specific rate of exponential fall-off of stellar
velocity dispersion (Rv = 2RD ) to make this possible. Here, Rv is the scale
length with which the vertical velocity dispersion falls exponentially with
radius. Recall that for a one-component, isothermal disk, z0 = vz2 /2πGΣ
(Eq.(3.5)), hence, z0 is constant when Rv = 2RD . Both these claims have
been accepted in the literature, although there was already observational
evidence by 1990s for a moderate increase with radius in the vertical scale
height [29, 126], and there is no compelling physical reason for the relation
Rv = 2RD to be valid.
In an important paper, [35] did a careful re-examination of the analysis
by [22] for two prototypical edge-on galaxies, NGC 891 and NGC 4565, and
showed that the data actually indicates a moderate increase of scale height
with radius, by a factor of ∼ 2 − 3, within the optical disk. Next, [35] applied
the multi-component disk plus halo model in the thin disk case, and showed
that the above variation in scale height can be used to constrain RV , which
is found to be > 2RD . The main points of this work are summarized next.
Re-examination of the analysis by van der Kruit & Searle
In the study of edge-on galaxies by [22], the intensity profile, I(R, z), is
measured along a cut normal to the edge-on disk at a given galactocentric
radius, R for the galaxy (as given in section 2.1.1). From these intensity pro-
files, [22] obtained a composite z profile by vertically shifting the individual
z profiles obtained at different radii and pinning them together at an arbi-

62
trary interim point z ′ . Consider the treatment for NGC 891 first. The above
procedure gives the top curve in Fig. 16. Next, the model by [22] for the
intensity of an edge-on galactic disk (Eq.(2.2)) gives a single and well-defined
model composite curve if the scale height, z0 , in Eq.(2.2) were constant at
all radii – this is shown as the solid curve in Fig. 16. [22] pin the data at a
point z ′ = 1.5 kpc, and claim that the single composite model curve (solid
curve) gives the best-fit to the observed data with little scatter around it.
[35], however, noted that the observed curves are not exactly in coincidence
into a single composite curve, as would be expected if z0 were constant with
R as claimed by [22]; instead, the data points show a considerable spread
around this single curve. [35] pointed out that the spread in observed data
around this composite curve corresponds to a spread in intensity, ∆I, at z=0
to be equal to ∼ 1 mag arc sec−2 .
Further, [35] made an important point that this spread being measured
on a logarithmic scale, could in reality indicate a substantial variation in
scale height with radius. They showed this quantitatively as follows. [35]
allowed z0 to vary with radius and found by trial and error that when z0 is
increased linearly from 0.75 kpc to 1.25 kpc between 0-20 kpc or the entire
optical disk in NGC 891;10 the resulting spread in the calculated values of
surface brightness for an edge-on disk (obtained using Eq.(2.2)), shown as
dots in Fig. 16, show a finite spread of intensity, ∆I ∼ 1 mag pc−2 at z=0
around the resulting composite profile. This exactly matches the spread in
the observed original data of [22] (shown within error bars, see Fig. 16).
Thus, the spread in the observed data actually allows for a linear increase
in z0 from 0.75 kpc to 1.25 kpc, or by a factor of 1.7, over the optical disk.
This is a substantial variation with radius compared to the strictly constant
z0 claimed by [22].
It should be stressed that this conclusion was reached by [35] using exactly
the same data and the model for luminosity for an edge-on disk as in [22].
The new point in the [35] study was that they noted the pertinent point that
the data actually shows a significant scatter of ∼ 1 mag arc sec−2 around
the composite curve, and this scatter being measured on a logarithmic scale
could indicate a substantial increase in scale height with radius over the
optical disk.

10
The radius of the optical disk is taken to be 4 RD in this work, see Appendix A for
details.

63
Figure 16: Plot of surface brightness, I vs. z for NGC 891 from [22] who used their data
and obtained this composite z-profile (the top curve in this figure) by vertically shifting
the individual z-profiles at different R into coincidence at z’= 1.5 kpc. The vertical bars
indicate the range of the data. The solid line is the model composite curve obtained by
[22] using Eq(2.2) where z0 = 1 kpc, taken to be independent of radius. [22] claim that
this model composite curve (solid line) fits the composite z-profile obtained from their
data. On the other hand, [35] show that, when z0 is allowed to increase linearly from
0.5 to 1.25 kpc within 20 kpc; the calculated surface brightness values (shown as points),
disperse over an interval of 1 mag arc sec−2 at the mid-plane. This is exactly as seen in
the original data of [22]. Thus, [35] show that the observed data can allow for as much
as a factor of 1.7 increase in scale height over the optical disk. The systematic deviation
of observed data at high z is due to the Thick disk which is not included in this study.
Source: Taken from [35]

64
Figure 17: Plot of resulting stellar scale height (HWHM) vs. R for NGC 891, calculated
using the multi-component disk plus halo model, for different values of rate of fall-off of
the vertical velocity dispersion, Rv . The variation in the scale height with R is found to
critically depend on the value of the ratio Rv /RD . The scale height is found to increase
by a factor of 1.8, 2.8 and 4.5 for the ratio Rv /RD = 2.5, 3 and 4, respectively, within the
optical disk. Thus, the scale height is not constant with radius. Source: Taken from [35]

Constraining the value of Rv /Rd

Next, [35] show that the above deduced variation in scale height can be
understood in terms of a disk in pressure equilibrium, and this variation
allows them to constrain the value of Rv . Due to the paucity of data for
velocity dispersion at that time [57], the observed value of Rv was not known.
[35] apply the multi-component disk plus halo model to NGC 891, and using
observed values of other input parameters for stars and gas, obtain the self-
consistent vertical distribution, and hence the vertical stellar scale height as
a function of R, for a range of values of Rv /RD covering 1 to 4 (see Fig. 17).
It should be pointed out that, the scale height obtained using the multi-
component disk plus halo model (given as HWHM), by definition, is not
identical to z0 – but the two are equivalent, and comparable, see Appendix
A. So the trends in variation in these two can be compared. This figure
shows that, the scale height typically increases by a factor of few within the
optical disk (for Rv /RD > 2). From the results in Fig. 17, [35] conclude that
the factor of 1.7 increase in the scale height as deduced from the spread in

65
the observed data for NGC 891, as discussed above, constrains the value of
Rv /RD to be between 2 and 2.5.
A similar analysis for another edge-on galaxy, NGC 4565, done by [35]
shows that the spread in the observed data indicates a linear increase in the
vertical scale height by a factor of 2.5 within the optical disk; which they
show constrains the range of Rv /RD to be between 2.5 and 3.
Interestingly, a small change in Rv /RD to 2.5-3 compared to the typical
value of 2 that is used, is adequate to explain the observed moderate flaring
of the stellar disk. In Section 5, we discuss the general physical significance
of the value of Rv /RD (> 2) and why it indicates a flaring disk.
Discussion:
Unfortunately, the paper by [35] and the important result from it, namely
that the vertical scale height is not constant; rather, it shows a moderate ra-
dial increase, has been largely overlooked in the literature. We note that the
recent observations and simulations also strongly support the above conclu-
sion of a flaring stellar disk (see Section 5).
Finally, a few more caveats about the analysis by [22] are given below.
These contribute to uncertainties in the determination of the variation of
scale height by [35]. First, the spread in the intensity profiles in the analysis
by [22] is controlled by the choice of the pinning point, z ′ , which is arbitrary.
The apparent spread can be minimized by an appropriate choice of z ′ . This
point was explicitly shown by [91], though it was not mentioned in the paper
by [22]. The actual choice of z ′ chosen would not matter if z0 were constant
at all R, since any choice of z ′ would then yield a complete overlap and hence
a single composite curve. Second, the spread in data is not just due to the
scale height variation but also due to other effects such as inclination, dust
extinction etc (see the discussion point 1 in [35]).
It would be worthwhile to look at this problem for a larger set of galaxies.
However, in view of the caveats in the analysis of [22] as described in this
subsection (and in Section 2.1.1), it may be advisable to follow a different
approach to see if the stellar disk is flaring. For example, z0 (R) could be
measured directly by an iterative analysis of intensity profiles as proposed
in Section 2. Alternatively, if the modern IFU data gives values of stellar
velocity dispersion directly, then using the multi-component disk plus halo
model; the scale height as a function of radius can be calculated. This would
be discussed in more detail in Section 5.4.

66
4.3. Multi-component disk plus halo model: Thick, or low density disk
Here we relax the assumption of a disk being thin and consider a thick
disk case. We next consider and compare the various cases systematically:
thin and thick disk cases – first a single component disk and then a multi-
component disk plus halo; and see how the formulation of equations for the
vertical structure is different in each case.
We stress that the term thick disk as used here and in the rest of the review
is based on its general physical characteristics alone, namely a disk with a
high physical thickness and/or a low density. In this case the R term in the
disk Poisson equation needs to be included in the formulation of equations
to obtain the correct disk vertical structure. In other words, the term thick
disk as used here or hereafter in this review does not specifically denote
the chemically and kinematically distinct structural component, which is a
common feature of many galaxies including the Milky Way. In this review,
the latter component will be refereed to as the Thick disk (e.g., Sections 1,
4.5.1, 7.1, 8). However, the results obtained in this section for a physically
thick disk are general and could be applied to study the vertical structure
of the structurally and chemically distinct Thick disks in galaxies (which is
outlined as a future problem, for details see Section 7.1 and Section 8).
As we have seen, to obtain a self-consistent vertical distribution in a
self-gravitating disk, the Poisson equation and the equation of hydrostatic
equilibrium have to be solved together. So far we have considered a thin disk
case: for a single component disk (Section 3.1); a multi-component disk in
the field of the halo as applied to the Milky Way, and also applied to model
two external galaxies, NGC 891 and NGC 4565 (see Section 4.2 for details).
In the above studies, the radial part of the disk Poisson equation is taken to
be zero.

4.3.1. Inclusion of radial term in the disk Poisson equation

Now let us consider the general case when the radial term in the Poisson
equation for the disk needs to be included in the formulation of the equi-
librium vertical disk distribution. This can happen either when the disk is
physically thick, or when the disk density is low as in the outer disk or at re-
gions away from the mid-plane, so that the term representing the self-gravity
in the Poisson equation, namely, (4πGρ0 ) is no longer much greater than the
R-term (see for example, Eq.(2-56) in [2] for an expression of Poisson equa-
tion for an axisymmetric disk). In other words, the disk can no longer be
taken to be locally dense. In this case, both, the radial and the vertical terms

67
in the disk Poisson equation, need to be included for a correct formulation
of the problem, see [20] for details.
As a physical consequence of including both the R and z terms in the
Poisson equation (in a thick or low density disk); the planar and vertical
dynamics in the disk are coupled. This is in contrast to the thin disk case
where the R term can be dropped and then the planar and vertical dynamics
are decoupled (Section 3.1).
Disk-alone case: Thick or a low density disk
Consider the disk-alone case first. The R term in the disk Poisson equa-
tion may be written in terms of the derivative of the rotation velocity, Vc ,
and is non-zero for a non-flat rotation curve. This ignores the cross terms
in the radial Jeans equation – this is justified for z < 1 in a galactic disk,
as discussed later, after Eq.(4.11), in this section. Thus the inclusion of the
R term changes the solution for the disk vertical distribution from the stan-
dard sech2 solution (Eq. 3.5) obtained for an isothermal, one-component
disk. This was shown for the Galaxy (see Model A in [20]) for which the
resulting mid-plane disk density, ρ0d , and HWHM are shown to be different
from the values for these for the standard one-component, isothermal sech2
solution by up to 10% in the inner Galaxy (where the R term is non-zero
since the rotation curve is not flat). It is easy to see how the dynamics along
R and z is coupled here via the Poisson equation. A more general treatment
for the R-z coupling was done for the one component disk-alone case by [20]
where the full Jeans equations were used that contained the cross terms and
also the additional terms involving the radial and azimuthal velocities (see
Section 4.4.1 for details).
Multi-component disk plus halo case: Formulation of equations for
a thick disk
Let us now compare how the joint Poisson equation behaves in the thin
and thick disk cases for the multi-component disk plus halo system. Recall
that for a thin disk, only the z term of the Poisson equation for the disk
needs to be retained (Section 3.1). In this case, the halo drops out of the
superposition or the joint Poisson equation for the disk and the halo, as
discussed in Section 4.2.1. The joint Poisson equation then reduces to that for
the disk-alone case, namely Eq.(4.3), where ρ denotes the redistributed disk
density. This redistribution occurs because the two structural components;
namely, the disk and the dark matter halo, are still gravitationally coupled,
through the equation of hydrostatic equilibrium (see Eq. (4.1), Section 4.2.1).

68
 Next, let us consider formulation of equations for the disk vertical struc-
ture of a thick or a low density disk. In contrast to the thin disk case, here
the R term in the disk Poisson equation needs to be included for a correct
formulation of the equations. In this case, the joint Poisson equation for the
gravitationally coupled disk and halo system is given by:

∂ 2 Φtotal
 
1 ∂ ∂Φtotal
R + = 4πG (ρs + ρHI + ρH2 + ρh ) (4.8)
R ∂R ∂R ∂z 2

where Φtotal = Φs +ΦHI +ΦH2 +Φh is the total potential, where the first three
terms are due to the three disk components considered (as defined in Section
4.2). In this case, the disk and halo together give the net gravitational force
that keeps the disk in a rotational equilibrium. In this case, the net radial
term in the joint Poisson equation is written in terms of the radial derivative
of the net circular velocity, Vc , at the point R under consideration, as follows:

1 ∂Vc 2
 
1 ∂ ∂Φtotal
R = (4.9)
R ∂R ∂R R ∂R
where Vc includes the contribution of disk and halo. Thus, in the thick disk
case, the Poisson equation for the disk and halo remain coupled through the
radial, and also, the z terms (see Eq.(4.8)).
We stress that Vc as defined above is taken to be equal to the observed
rotational velocity, for simplicity. In writing the disk contribution to the
R term of the Poisson equation in Eq.(4.9), the contribution of the cross
terms and the velocity dispersion terms in the radial Jeans equation (Eq.
4.29 a from [2]) is neglected. We note that this is equivalent to neglecting
the asymmetric drift (e.g., [2]). This point is discussed in detail later, after
Eq.(4.11). Hence Vc as defined by Eq. (4.9) is taken to denote the observed
rotational velocity in the rest of the review; with the exception of Section
4.4.1, where the complete Jeans equations are used to write the R and z
derivatives of the potential.
Another point to note is that, strictly speaking, Vc is a function of both R
and z; and is specified as Vc (R, z). However, typically, the observed rotation
velocity at each R is given as the intensity-weighted average of Vc (R, z) along
the vertical direction, which is a function of R alone. Thus, usually for sim-
plicity, the rotation velocity at a given R, Vc (R), is taken to be independent
of z. Therefore, the term on the r.h.s. of Eq.(4.9) is a function of R only,

69
and its value is determined from the gradient of the observed rotation curve.
The general case where the rotation velocity, Vc is a function of z will be
considered later in this section.
The joint Poisson equation in terms of the observed rotation velocity is
obtained by substituting Eq.(4.9) into Eq.(4.8), and is given to be:

∂ 2 Φtotal 1 ∂Vc 2
( ) + = 4πG(ρs + ρHI + ρH2 + ρh ) (4.10)
∂z2 R ∂R
It is interesting that the rotation curve features in the formulation of the
equations in the thick disk case. In contrast, for the thin disk case, the R
term or effectively the rotation curve does not feature in the formulation of
the equations. This is true, both, for the one-component disk-alone case (see
Section 3.1) because the R term in the disk Poisson equation drops out; and
also for the multi-component disk plus halo thin disk case (Section 4.2.1),
because the disk and halo Poisson equations are effectively decoupled. Hence
the halo R term does not feature in the formulation of the equations for the
vertical density distribution in the latter case.
In a multi-component galactic disk plus halo system, the gravitational
force due to the other disk components as well as that due to the halo, is
included in the equation of hydrostatic balance for each disk component.
This can re-arrange or modify the ρ and Φ of the particular disk component
(Section 4.2).
The equation of hydrostatic balance for each disk component in the cou-
pled multi-component disk plus halo model remains the same as in the thin
disk case, namely:
∂ ∂Φtotal
(ρi ⟨(vz2 )i ⟩) + ρi =0 (4.11)
∂z ∂z
where the various quantities are as defined for the thin disk case (4.2.1).
While writing these expressions (Eq.(4.9), Eq.(4.10), and Eq.(4.11); truly
speaking, the radial and z derivative of the potential should be given in
terms of the full Jeans equations (e.g., Eq. 4.29a and Eq. 4.29c respectively,
from [2]). The latter also include terms containing the cross terms and the
velocity dispersion, which have been dropped here, for simplicity. The ratio
of the terms containing the cross terms and the velocity dispersion (which
are dropped) and the other terms (which are kept) is ∼ z 2 /RRD , which is
∼ 0.01 so long as z < 1 kpc. This has been shown for the radial Jeans equation
[5, 86, 89]; and, also the vertical Jeans equation ([86, 2, 91]; also, see Section

70
3.1). Recall that, here the radial term in the disk Poisson equation has been
included so as to treat a low density or a thick disk. But, interestingly,
even when the disk is thick; so long as z < 1 kpc, the cross terms can be
still ignored from the radial and vertical Jeans equations, as shown above.
Hence the above simplified forms of the Poisson equation (Eq.(4.10)) and
the equation of hydrostatic balance used (Eq.(4.11)) for a thick disk case are
valid for typical real galactic disks (since these have a thickness < 1 kpc).
Dropping these terms as done for the R term of the disk Poisson equation
(Eq. 4.29 a from [2]) is equivalent to neglecting the asymmetric drift. Hence,
we stress again that Vc in Eq. (4.9) denotes the observed rotation velocity.
However, as shown in a general model ([20]; also, Section 4.4.1), the
inclusion of the cross terms and the terms involving velocity dispersion, and
the tilt of the velocity ellipsoid; makes a significant difference at large R, or
in very low density or high z region. In these regimes, these additional terms
need to be included for the correct formulation of the equations.
On combining Eq. (4.10) and Eq.(4.11), the coupled, joint Poisson-
hydrostatic balance equation for the multi-component disk plus halo system,
for the thick disk case, is obtained to be:

1 ∂Vc2
 
∂ 1 ∂ρi
⟨(vz2 )i ⟩ = −4πG (ρs + ρHI + +ρH2 + ρh ) + (4.12)
∂z ρi ∂z R ∂R
The solution of this coupled set of equations gives the self-consistent ver-
tical density distribution for the thick disk case, for each of the coupled disk
components (i= stars, HI and H2 , respectively), with the disk being under
the gravitational field of the halo. The numerical solution for these is ob-
tained following the same procedure, including the boundary conditions at
the mid-plane; as done for the multi-component, thin disk (as in [16]; also,
see Section 4.2.2, and Eq.(4.5)).
One-dimensional, local calculation
A somewhat subtle technical point is that despite the inclusion of the
radial term, one can still treat the surface density, Σ(R), to be constant at
a given R to obtain (ρ0 )i , the modified mid-plane density. This is, in fact,
one of the boundary conditions (see Eq.(4.5)) while obtaining the numerical
solution to Eq. (4.12). For a disk in rotational equilibrium, the net R term
in the Poisson equation is a constant at a given R and its value depends
on the gradient of Vc at R alone (although the rotational equilibrium takes
into account the mass distribution at non-local regions inside of R.) Thus,

71
even though the R and the z terms in the Poisson equation are coupled,
the inclusion of the radial term only changes the effective gravity, and hence
the z motion, locally at a given R. Thus, despite the inclusion of the radial
term, the surface density at a given R remains unchanged. Hence, the con-
stant surface density can still be used as a constrain to determine one of the
boundary conditions, namely, (ρ0 )i . The other boundary condition remains
as dρ/dz = 0 at the mid-plane (or, z = 0), as before ([16]; also, Section
4.2.2).
Here too, the problem reduces to a one-dimensional, local one of the de-
termination of ρ along z; as in the thin disk case (see section 4.2.1). Hence,
in the thick disk case, the vertical density distribution of a given disk compo-
nent in the coupled case is only affected by the local values of the parameters
of the other disk components, at a given R – as was also shown to be true
for the thin disk case (see Section 4.2.1).
Effect of non-zero rotation velocity on scale height
Note that the presence of the last term on the r.h.s of Eq.(4.12) opposes
the effect of the net gravitational force due to the local mass density if the
velocity gradient is positive. Thus, in the rising part of the rotation curve, the
last term opposes the disk gravity and this results in a puffed up distribution
or a higher scale height. In contrast, in a region of falling rotation curve, the
opposite is true; namely, the inclusion of the last term adds to the self-gravity
term. Hence, it reduces the resulting scale height. This will be illustrated by
example in case of real galaxies (see Section 4.3.2).
Application to galaxies with a rotational lag:
A careful study by various authors has revealed that there is a variation
with z in the rotation velocity in some galaxies; also known as the rotational
lag, with a lower rotational velocity at higher z values (e.g., [132, 133, 134,
135, 136, 137]). Here the gas rotational velocity has a negative vertical gra-
dient with z. In this case, Eq.(4.9) is modified as follows. Assume that the
lag in the rotational velocity can be specified as Vc (R, z) = Vc (R, 0) + Clag z;
where Clag = dVc /dz (< 0) is the linear decrease, or lag in the rotation ve-
locity with z. Further, if the lag is the same at all radii, that is, dClag /dR = 0;
then the r.h.s. of Eq.(4.9) is modified and contains one extra term, 2(z/R)Clag (dVc /dR).
The corresponding modified coupled, joint Poisson-hydrostatic balance equa-
tion (Eq.(4.12)) can be solved to obtain the resulting density distribution. It
can be shown that the typical observed lag of a few km s−1 kpc−1 , results in

72
very small (< few %) change in the vertical density distribution.11 However,
we point out that this discussion is based on the observations for HI gas and
ionized gas. It is not clear from observations whether the stars obey any
extra-planar rotational lag, and whether its magnitude is the same as that
observed for the HI gas.

4.3.2. Thick, or low density disk case: Applications

The above formulation of the multi-component, thick disk plus halo case
was general. In this subsection, it is applied to three specific cases: namely,
the Milky Way outer disk; the dwarf galaxies; and a low surface brightness
galaxy, UGC 7321. These are physically distinct cases and cover important
situations of interest in real galaxies. The details of the equations are slightly
different in each case due to the different physical conditions in these three
systems (e.g., the flat rotation curve in the Milky Way). Further, the results
obtained cover different aspects studied in each case (e.g., halo profile in the
Milky Way). These three cases are presented next.
Application to the Milky Way outer disk: Halo density profile
The formulation involving a thick disk for the multi-component disk plus
model was studied for the outer disk of Milky Way by Narayan et al (2005)
[78], to obtain the HI gas scale heights. The aim of this paper was to use the
observed HI scale heights and the observed rotation curve as two independent
constraints to determine the best-fit halo parameters. This approach will be
discussed in more detail in Section 6. This study involves one of the early
applications of the general thick disk case, as developed in Section 4.3.1 (Eq.
(4.12)). Here, the net radial term in the Poisson equation due to the disk
and halo is zero (Eq.(4.9)), because the observed rotation curve in the Milky
Way is flat (see [78] for details).
Recall that for the thin disk, on the other hand, the radial term in the
disk Poisson equation is zero ([16], Section 4.2.1). In that case, therefore, the
disk and halo Poisson equations are decoupled, as discussed earlier (Section
4.2.1); while in the thick disk case, even when the R term is zero, as for a
flat rotation curve; the disk and the halo Poisson equations remain coupled
(see Eq. (4.10)). Therefore, the resulting coupled, joint Poisson-hydrostatic
balance equations (Eqs. 4 and 5 in [78]), and the resulting solutions, are
different in these two cases. The thin disk assumption would overestimate

11
S. Sarkar, private communication. I thank S. Sarkar for checking this.

73
the scale height by as much as 10%. Since the resulting HI scale heights
are used to constrain the halo parameters, it is important to obtain accurate
model values for the scale heights. Hence [78] use the general approach that
considers a thick disk.
For a thick disk, and a flat rotation curve, the net coupled, Joint Poisson-
hydrostatic balance equations for a thick disk, Eq. (4.12), reduce to:
 
2 ∂ 1 ∂ρi
⟨(vz )i ⟩ = −4πG (ρs + ρHI + +ρH2 + ρh ) (4.13)
∂z ρi ∂z
For the given set of input parameters for the Galaxy, the solution of
these three coupled equations is obtained numerically using the same iterative
procedure and the boundary conditions as in [16]; also, see section 4.2.2. This
gives the self-consistent density distribution ρi (z), and the corresponding
HWHM denoting the thickness or vertical scale height at a given radius for
each disk component. These results are compared with the observed HI scale
height values.
To obtain the best-fit halo parameters, [78] try a large range of values for
the halo parameters while solving Eq. (4.13). They assume a four-parameter
halo model described by the following density profile [138]:

ρ0h (q)
ρ(R, z) = 2 (4.14)
(1 + Rm2 (q) )p
c

where ρ0h is the central halo density, Rc (q) is the core radius, q is the axis
ratio and p is the index. By definition, m2 = R2 + z 2 /q 2 represents the
surfaces of concentric ellipsoids. Note that q = 1 gives a spherical halo, while
q = c/a < 1 and q = c/a > 1 gives an oblate and a prolate halo respectively.
Here a and c are the semi-major axis in the disk plane, and that along the
vertical direction, respectively. Note that for p = 1 and q = 1, Eq. (4.14)
reduces to a pseudo-isothermal density profile (Eq. (4.6)).
By varying the density index p one can get different halo density profiles.
The choice p = 1 corresponds to a screened isothermal or a pseudo-isothermal
halo, with a density falling off as r−2 at large radii (r >> Rc ), this results
in a flat rotation curve at large radii. Here r is the radius in spherical co-
ordinates. For p = 1.5 and p = 2 the density falls off more steeply: as r−3 as
in the NFW (Navarro, Frenk & White) case[139]; and as r−4 , respectively.
At large radii, the mass for p = 2 to a finite value unlike the other two cases.

74
Figure 18: Plot of calculated HI scale height (HWHM) vs. R in the outer Galaxy. The
shape of an initially spherical isothermal halo (with ρ0 = 0.035 M⊙ pc−3 , and Rc = 5
kpc) is changed keeping its mass constant. The solid line results for the spherical shape
(q = c/a= 1); the dashed lines are for oblate halos (c/a = 0.8, 0.6, 0.4), and the dotted
lines are for prolate halos (a/c = 0.8, 0.6, 0.4). The points show the observed values.
Note that neither the oblate nor the prolate-shaped halos are clearly favoured by the data.
Beyond R= 20 kpc, a prolate halo is favoured, but no single value of prolate shape fits the
data over the entire radial range. Source: Taken from [78]

First, different shapes for the isothermal (p = 1) case are tried. On

varying the shape, q, with p = 1, the values of the parameters (ρ0h (q), Rc (q))
to be used in Eq.(4.12) change and can be obtained in terms of the spherical
case values, by imposing two physical conditions: namely, that the mass
within a thin spheroidal shell, and the terminal velocity of the halo should
be independent of q (see Eq.7, and Fig. 1 in [78]). For the fiducial spherical
halo, the pseudo-isothermal model (Eq. (4.6)) is used for which Rc (1) = 5
kpc and ρ0h (1) = 0.35 M⊙ pc−3 (as given by [82]).
For a given radius R, assuming a certain density profile for the halo
(that is, for a choice of p and q), the corresponding ρ0h (q) and Rc (q) are
obtained as described above. Further, using the input parameters for stars
and gas from observations (as given in Section 4.2.3); the solution for the
three coupled equations (Eq.(4.13)) yields the self-consistent density ρi (z),

75
and the HWHM, for each disk component. Repeating the same procedure
at different radii gives the model scale height curve. The HI scale height
results depend sensitively on the choice of the HI velocity dispersion (see the
discussion at the end of Section 4.2.4). A detailed discussion is given in [78]
regarding the choice of the HI dispersion used and its variation with radius.
The resulting HI scale height curve is shown for the outer region of the
Galaxy in Fig. 18. The solid line is due to spherical shape (q = 1), the
dashed lines are due to oblate halos, and the dotted lines are due to prolate
halos. The observed HI scale height values taken from [42] – given as points
in the figure – are used to fit the results from the model. The [42] data do
not have error bars with the data points. Hence, to estimate a goodness of
fit, [78] compute least-square of the model generated curve. Note that the
solid line corresponding to the scale height curve for a spherical halo gives a
good agreement with observations up to about R=20 kpc. Beyond that the
calculated values fall below the observed values, thereby giving a poor fit.
If the halo shape is changed to an oblate shape while keeping the mass
constant, physically it is easy to see that the gravitational force normal to
the plane will be higher and hence will result in a smaller HI scale height.
This would make the fit worse at all radii, and especially at large radii; since
the HI thickness is observed to steeply flare with radius in the outer disk,
beyond 20 kpc. Conversely, a prolate halo would give a better fit. However,
given the steeply flaring HI disk, no single prolate shape is found to give a
good fit over the entire radial range considered (see Fig. 18). A halo that
is progressively more prolate with radius could explain the observed sharp
flaring of HI scale heights in the outer Galaxy, as shown by [140]. This will
be discussed later in Section 6.3.1.
Next, [78] tried the halo density profiles falling steeper than r−2 in the
outer region and showed that these provide a better fit to the observed HI
flaring, and these also fit the observed rotation curve of our Galaxy. To check
this, density profiles for index values of p = 1.5 and p = 2 were tried, while
the shape was kept spherical for simplicity, and also so as to isolate the effect
of density variation of the halo. For each value of p, a large range of realistic
central density, ρ0h , and core radius, Rc , values were chosen to form a grid
of (ρ0h , Rc ) values. In each case, the model rotation velocity and the HI gas
scale height were obtained at each R. This was repeated for the entire range of
R values. By simultaneously comparing the model results with the observed
rotation curve and the gas scale heights following the detailed procedure as
given in [78], a good fit to the observed data was found for halos with the

76
Figure 19: Plot of HI scale height vs.R in the outer Galaxy. The best fit to the data
(shown as points) is obtained for a model with a halo with p = 2 (solid line) where p is
the index in the halo density profile. For comparison, results for a model with typical
isothermal halo (p = 1) are also shown (dashed line). See the text for details. Source:
Taken from [78]

central density and the core radius in the range of ρ0h = 0.035 − 0.06M⊙ pc−3
and Rc = 8 − 9.5 kpc, respectively; with p = 2 providing a better fit. Of
this range, the best fit was obtained for p = 2, and ρ0 = 0.035M⊙ pc−3
and Rc = 9.4 kpc, see Fig. 19. Recall that at large radii (>> Rc ), p = 2
corresponds to a density falling off as r−4 . This steeper fall-off helps explain
the steep flaring in HI scale height with radius in the outer Galaxy, between
R = 20-24 kpc. This density profile gives rise to ”truncated” or ”finite-sized”
halos with about 95% of the total mass within a few 100 kpc. These results
were argued to be in a good agreement with the observed data (e.g., from
SDSS), and numerical simulations of cosmological evolution of galaxies, as
available at that time.
We point out that given the tremendous growth in the observational data
as well as studies of numerical simulations, this subject is ripe for a re-look,
while adopting the same formalism as in [78].
Application to Dwarf galaxies:
Another example of an application of the multi-component disk plus halo
model in the thick disk case is the study of dwarf galaxies by [141]. The

77
aim of this study was to theoretically obtain the vertical scale height of
atomic hydrogen gas (HI) in four dwarf galaxies, DDO154, HoII, IC 2574, and
NGC2366. The dwarf irregular galaxies are low-mass, low-metallicity, gas-
rich objects and constitute the largest number of galaxies in the present-day
day observable universe [142]. These are of interest for early galaxy evolution
in the paradigm of hierarchical galaxy formation. These are believed to have
large HI scale heights [143]. The actual value of the gas scale height has
implications for different topics: including star formation, and the growth of
HI holes in these galaxies. Hence, to study the vertical density distribution;
the general, thick disk case is applied here Eq.(4.12). The treatment as a thick
disk serendipitously allows for the inclusion of the effect of the radial variation
of the rotation curve with R (see the related discussion in Section 4.3.1). In
contrast, if the disk is taken to be thin, the rotation curve does not feature
in the formulation of the equations (see the discussion after Eq.(4.10)).
In dwarf galaxies, over most of the observed region, the rotation curve
is seen to be rising, hence the radial gradient of observed rotational veloc-
ity is positive. Thus, the last term on the r.h.s. of Eq.(4.12) opposes the
self-gravity term. Therefore, the inclusion of the radial term here results in
slightly higher disk scale heights than in the absence of it, as already ex-
plained earlier in Section 4.3.1. The molecular gas content is negligible in
dwarf galaxies. Hence, the coupled, joint Poisson-hydrostatic balance equa-
tion, as given by Eq.(4.12) is applied to a two-component disk; with i= 1,
2 corresponding to stars and HI, respectively. These two joint second-order
coupled differential equations for stars and HI are solved simultaneously; fol-
lowing the same numerical procedure, including the boundary conditions,
as in [16]. The resulting solution gives the self-consistent vertical density
distribution for each disk component.
To do this, the observed input parameters for stars and gas are used (see
[141] for details). To solve Eq.(4.12), the dark matter halo parameters are
also needed. The halo is assumed to be spherical with a pseudo-isothermal
density profile (Eq. (4.6). The values of the parameters describing it, namely
the central density ρ0h and the core radius, Rc , were obtained in the literature
by fitting the model rotation curve to the observed rotation curve. Using the
above input parameters, the self-consistent solution for the vertical density
distribution, ρ(z), and the associated HWHM, were obtained for HI gas. This
process is repeated at different radii to get the resulting scale height variation
with radius.
Interestingly, this study employs an inverse approach to the one taken in

78
Figure 20: Plot of results for the HI scale height (HWHM) vs. R for two dwarf galaxies
studied: DDO 154 and NGC 2366. The solid line and a dashed line in each case correspond
to the thick disk treatment: that includes the variation of rotation curve with radius, and
one that assumes a flat rotation curve, respectively. The two show a small difference – see
the text for details. The scale height increases steadily with radius up to the last point
for which data are available. Source: Taken from [141]

[78] to study the Milky Way – as seen earlier in this subsection. In [78], the
observed HI gas scale height values were used to constrain the best-fit halo
parameters, using the same basic model – except there the observed rotation
curve and HI scale heights were used as two independent constraints to obtain
the halo parameters.
Fig. 20 shows the resulting HI scale height (HWHM) as a function of R for
two of the galaxies studied, DDO 154 and NGC 2366. In each case, the results
are shown as a solid line and a dashed line: both of which correspond to the
thick disk treatment, one that includes the variation of rotation curve with
radius, and the other assuming a flat rotation curve, respectively (obtained
using Eq.(4.12) and Eq.(4.13), respectively). In each case, the scale height
increases steadily with radius up to the last point for which data are available,
thus flaring by a factor of few within several disk scale lengths. Since the
rotation curve is rising over most of the disk, the resulting scale height values
for the thick disk treatment which includes the variation in the rotation curve

79
are slightly higher (by ∼ 10 − 20%), than those obtained assuming a flat
rotation curve, as expected (see the discussion above).
The values of HI scale height obtained are in the range of ∼ 200 − 400
pc in the inner disk region, increasing up to ∼ 600 pc-1 kpc out to the last
measured point. The inclusion of gas gravity and the gravitational effect of
halo as done here gives more reliable results for the scale heights - with smaller
resulting values, than obtained in the previous studies in the literature (which
ignored gas gravity). This is important because the gas scale height value,
in turn, is used for deriving other physical quantities, as discussed above.
Measuring the gas scale heights in galaxies is difficult: due to possible
contamination by flaring and warps ([144, 145, 42]; and, due to inclination
effects [144], also see Section 6.5). As an independent check, [141] compared
the derived scale heights with the observed size distribution of type 3 HI
holes (roughly spherical cavities that are contained within the gas layer),
and found these to be in a good agreement. This is an indirect confirmation
of the model scale heights obtained for these galaxies.
However, it should be pointed out that dwarf irregular galaxies show a
large variation of parameters [146]. Hence, while the results for the HI gas
scale height from this paper can be taken as clear trends, these need to be
confirmed for a larger sample of irregular galaxies in a follow-up work.
Application to a LSB galaxy, UGC 7321:
An interesting application of Eq.(4.12) (which was meant for a thick or a
low density disk) is to an edge-on low surface brightness (LSB) galaxy, UGC
7321 [91]. The LSBs form a special class of galaxies that lie at the faint end
of the galaxy luminosity function, and are unevolved. LSBs have a low mass
disk, and a dark matter halo that dominates from the innermost regions.
Thus the LSBs are structurally, and hence dynamically, different from the
typical high surface brightness (HSB) galaxies, like the Milky Way. Hence
it is of interest to study the vertical distribution of the stellar disk of a LSB
galaxy and see how it compares with, say, that of the Milky Way disk.
Interestingly, here, in absolute terms the disk is thin due to the constrain-
ing effect of the dominant halo [147]. Yet, because of the low disk mass and
density, the term denoting self-gravity in the disk Poisson equation does not
dominate over the R term. Hence, the R term in the disk Poisson equa-
tion has to be included to correctly formulate the problem of the vertical
density distribution in the disk (Section 4.3.1). Hence, here too, the self-
consistent vertical density distribution is determined by the coupled, joint

80
Poisson-hydrostatic balance equation given by Eq.(4.12).
UGC 7321 was chosen for this detailed study to obtain the stellar scale
heights because it is one of the LSBs that has been well-studied; so that the
various input parameters, such as the stellar and HI gas surface density, and
the HI velocity dispersion, were known observationally. The central stellar
velocity dispersion value has been estimated assuming the stellar disk to be
in pressure equilibrium [148]; see [91] for details of the parameters. Further,
this is one of the few LSB galaxies whose vertical stellar structure has been
studied observationally. Its vertical scale height has been measured up to
R= 5.6 kpc, and it is found to be increasing with radius [148]. Here, the disk
scale length, RD is 2.1 kpc [149].
The dark matter halo is assumed to have a pseudo-isothermal density
profile (Eq.(4.6)). The net rotation curve obtained by quadratically adding
the stellar disk and halo contributions to it, is fitted with the observed ro-
tation curve from [150]12 . The best-fit halo parameters are obtained to be
ρ0h = 0.126M⊙ pc−3 and Rc = 1.4 kpc, where ρ0h and Rc are the central
density and the core radius of the halo, respectively. These values indicate
a dense and compact halo (as defined by [147]). The molecular gas content
in LSBs is negligible. Using the above parameters, Eq.(4.12) is solved for
a two-component case to obtain the self-consistent vertical density distribu-
tion for stars and HI. To do this, the same numerical procedure, including
the boundary conditions, as in [16] is used.
Flaring stellar disk
The value of Rv , the rate at which the stellar velocity dispersion falls
exponentially with radius, is obtained by fitting the resulting model scale
height values to the observed HWHM vs. R for stars (known up to R=5.6
pc) used as a constraint. This procedure gives the best-fit value for Rv =
3.2 RD [91]. The same rate of fall-off is assumed to hold good for radii
beyond 5.6 kpc until R=8 kpc when the stellar velocity dispersion falls to
the gas dispersion value. Beyond this radius, the stellar velocity dispersion is
assumed to saturate because of the physically motivated argument (see [19]),
that the stellar dispersion cannot be less than the dispersion in the gas from
which the stars form. The model scale height is obtained using the above
velocity dispersion values as input. The resulting scale height continues to

12
L.D. Matthews, personal communication (2009). I thank L.D. Matthews for tabular
form of this data.

81
increase beyond 5.6 kpc, the increase is gradual up to 7 kpc and then it flares
rapidly (see Fig. 5 from [91]). The stellar disk thickness increases (from ∼
200 pc to 680 pc) or by a factor of ∼ 3 from R = 0 to 10 kpc (see Fig. 5
from [91]). The resulting HI disk thickness increases (from 130 pc to 820 pc);
that is, by a factor of 6 from R= 0 to 10 kpc (see Fig. 6 from [91]), in good
agreement with observations.
[91] note that the best-fit value of Rv = 3.2RD ) obtained here is at vari-
ance with the choice of (Rv = 2RD ) routinely used in the literature. The
latter was proposed by [22]; also, see Section 2. Therefore, [91] emphasise
that the choice of Rv = 2RD is not physically justified in dynamical model-
ing of galaxies, and could give erroneous results. This point, along with the
relation between the value of rate of fall-off of velocity dispersion, or Rv /RD ,
to flaring, is discussed in detail in Section 5.
This work clearly shows that although UGC 7321 is dominated by dark
matter halo at all radii, the stellar disk shows flaring. Thus, a flaring stellar
disk appears to be a generic phenomenon in LSB galaxies as well.
Density profile of stellar disk and wings at high z
Next, [91] fit the results from their theoretical model for the self-consistent
vertical density distribution for UGC 7321 at different radii, to the function
proposed by [14], (Eq.(4.7). The best-fit value for the parameter (2/n) is
obtained; first, as a function of R (over the interval taken for fitting, ∆ z=
200 pc); and then as a function of ∆ z at a given R. It is found that the
best-fit value of 2/n fitted over ∆z = 200 pc changes with R and starts to be
close to 2 or greater than 2 (corresponding to n < 1) from R= 5 kpc (or, ∼ 2
RD ) onward; that is, in the outer disk. This region is entirely dominated by
dark matter halo gravity. A similar trend was found for the Galaxy beyond
R=18 kpc by [19], who pointed out that this range of n is new and was not
considered earlier in the literature (see Section 4.2.6). This range of (n < 1)
corresponds to a broad distribution at high z. Further, at any radius, the
exponent 2/n increases with increasing ∆z values; so that, a single n does
not provide a good fit over the entire range of ∆z considered. Thus n is not
a robust parameter. This result was also seen for the Galaxy [19]. At large
R > 5 kpc and high z, n is < 1 (see Fig. 8 from [91]). Thus, at large R and
z, the stellar distribution would be broader than sech2 and would appear as
broad wings in the intensity profiles.
Observations show that in UGC 7321 there is an excess emission or wings
at high z, that deviate from the distribution at low z. To explain this, a

82
second disk, with a higher scale height was invoked earlier [148].
However, as discussed above, [91] showed that a physically motivated
multi-component (single) disk plus halo model naturally gives a broader dis-
tribution at high z and high R; without the necessity of invoking a second,
thicker disk as done by [148] for UGC 7321; and also for another LSB galaxy,
FGC 1540, by [151]. [91] therefore argue that such a second, thicker disk
would, in fact, be redundant.

4.4. General models

In this section, we discuss more general models, which allow one to explore
even more realistic cases. First is a general model for a thick, isothermal,one-
component disk using the complete Jeans equations which effectively includes
various kinematical effects. The second is the thin, non-isothermal, multi-
component case in the field of the halo. In both, the idea was to to highlight
the effect of a special new physics point considered, while keeping the problem
tractable. Both these points are shown to significantly affect the vertical
density distribution in a galactic disk.

4.4.1. Model for thick, or low density disk, using complete Jeans equations
Sarkar & Jog (2020a)[20] consider the general, complete model for an
isothermal, one-component, thick or low density disk, that includes both the
radial and vertical terms in the Poisson equation. These are then written
in terms of the full radial and vertical Jeans equations which take account
of the non-flat observed rotation curve, the random motions, and the cross
term that indicates the tilted stellar velocity ellipsoid. This is a more general
and complete treatment for a thick disk than given in Section 4.3; where the
radial term in the disk Poisson equation was taken into account, and this
term was expressed in terms of the rotational velocity only.
This complete approach is applied to the Milky Way, and it is found that
these additional kinematical effects result in a density distribution that is
significantly different from the standard sech2 law.
The cross term vR vz where the average is taken over the velocity disper-
sion, is a component of the velocity ellipsoid tensor whose value is set by
the tilt of the velocity ellipsoid w.r.t. the disc plane. It also indicates the
coupling between the radial and vertical motions. The terms containing the
cross terms are smaller than the other terms in the R and Z Jeans equations
by a factor of ∼ z 2 /(RRD ), which has a value of a few % for z < 1 kpc in

83
the solar neighbourhood ([89, 2]; and the discussion in Section 4.3.1). Hence,
typically, these terms have been neglected in dynamical studies for a thin
disk – this is justified near the mid-plane. However, recent kinematic data
from various observations such as RAVE (Radial Velocity Experiment), SDSS
(Sloane digital sky survey), and Gaia shows that the stellar velocity ellipsoid
is tilted in the meridional plane in the stellar disk [100] and is observed to
have an orientation such that it aligns with the spherical polar co-ordinate
system centred at the center of the Galaxy. Motivated by this, [20] included
the cross terms; and to cover all aspects, also included the other terms in the
Jeans equations. The inclusion of all these terms is shown to significantly
affect the vertical density distribution. Some earlier studies had used the full
Jeans equations to determine: the vertical force field, the shape of the dark
matter halo, and local estimate of dark matter density [99, 100, 152, 153];
but the effect on the vertical distribution of stars had not been studied.
The formulation of the equations in [20] is briefly described next. The
complete Poisson equation is used that includes both radial and vertical
terms, which is given by:

1 ∂ ∂Φ ∂ 2Φ
(R ) + 2 = 4πGρ(R, z) (4.15)
R ∂R ∂R ∂z
To calculate the radial and the vertical gradient of the potential, the
complete radial and vertical Jeans equations are used, given as follows (from
[2]):
2
∂Φ 1 ∂ 2 1 ∂ (vR − vϕ2 )
=− (ρvR )− (ρvR vz ) − (4.16)
∂R ρ ∂R ρ ∂z R
∂Φ 1 ∂ 1 ∂
=− (RρvR vz ) − (ρvz2 ) (4.17)
∂z ρR ∂R ρ ∂z
Here vR , vϕ , vz denote velocities along the three axes. Assuming that
there is no net streaming motion along radial and vertical directions in the
Galaxy, one can write vR2 = σR2 and vz2 = σz2 where σR , σz represent velocity
dispersions of stars along R and z directions respectively.13 The quantity vϕ2

13
In the rest of the review, the quantity ⟨(vR2 )⟩ (=σR2
) has been used to denote the
2 = σ 2 is used
mean square velocity dispersion along R. In this subsection, the notation vR R
to denote this same quantity. This is for convenience, so as to be able to use the same

84
can be written as vϕ2 = σϕ2 + vϕ 2 , where vϕ represents observed mean rotation
velocity of stars in the disk and σϕ is the random velocity dispersion along
the azimuthal direction.
When the velocity ellipsoid is aligned with the spherical co-ordinate axes
centred at the centre of the Galaxy, then the cross term vR vz is given as
(σR2 − σz2 )(z/R) [86, 2]. Substituting these in Jeans equations, the latter
are simplified. Further, as usual, the surface density and the radial stellar
velocity dispersion are assumed to fall exponentially with radius, with a scale
length RD and Rv , respectively (Section 2). The ratio of the vertical and
azimuthal dispersions in terms of the radial velocity dispersion is defined as
bz = σz2 /σR2 and bϕ = σϕ2 /σR2 , respectively, and these are obtained in terms of
observed quantities. These ratios are taken to be constants; which effectively
means that Rv , the rate of fall-off in velocity dispersion is assumed to be the
same for radial and vertical velocity dispersions (Section 4.2.3).
Using these assumptions and on substituting the Jeans equations in the
Poisson equation, gives the net second-order differential equation describing
the self-consistent vertical density distribution, ρ(z), to be:

 2
∂ 2ρ −4πGρ2 1 ∂ρ
 
ρ 2vϕ ∂vϕ
= + + 2
∂z 2 σz2 ρ ∂z σ R ∂R
  z
z(1 − bz ) 2 2 ∂ρ
− −
bR RD Rv ∂z
z (4.18)
ρ 2 4 1 2 4(1 − bz )
+ − 2+ − +
bz RRv Rv RRD RD Rv RRv

2(1 − bϕ ) (1 − bz )
+ +
RRv RRD
The resulting self-consistent density distribution along z can be obtained
by solving this equation at a given R; following the numerical procedure as
in [16]. Next, the radial variation of ρ(z) is obtained.
It should be pointed out that if only the first two terms on the r.h.s. of
Eq.(4.18) are considered; it reduces to the usual thin, isothermal case with
a solution proportional to sech2 . The third term on the r.h.s. arises due

notation for the equations formulated by [20], which are used here. Similar notation is
employed for the z case as well. In this subsection, the velocity dispersions are taken to
denote that for the stars (and are given without the subscript s).

85
to the radial term in the Poisson equation when expressed only in terms of
the observed rotation velocity (as in [91]). Thus the rest of the terms on
the r.h.s. of this equation denote the additional terms that arise due to the
inclusion of the terms involving velocity dispersion and cross term in the
radial and vertical Jeans equations. The three cases described above have
been named as the sech2 law, model A, and model B respectively. These allow
a systematic comparison of the most general, complete model considered here
(model B) with the other two cases. For the detailed results for these cases,
see Table 1 in [20] – a few highlights are given next.
Results:
Consider case A first. The joint Poisson equation-Hydrostatic balance
equation for model A is given as follows:
 2
∂ 2ρ −4πGρ2 1 ∂ρ
 
ρ 2vϕ ∂vϕ
= + + 2 (4.19)
∂z2 σz2 ρ ∂z σz R ∂R
The rotation curve, vϕ , is taken to be flat at all radii starting from 8.5 kpc
and beyond. On comparing the results from model A with the sech2 case, the
two are found to differ by ∼ 10% upto R < 8.5 kpc, and are identical beyond
that due to the assumption of a flat rotation curve. The difference at low
radii could be of either sign, which depends on the sign of the gradient of the
observed rotation velocity. As expected, a rising rotation curve effectively
reduces self-gravity (Section 4.3.1, see the discussion after Eq.(4.12)); and
hence, gives slightly puffed up disks with a lower mid-plane density. Hence,
the difference between results from Model A, which has a rising rotation
curve, and sech2 , differ by a few % value (see Table 1 in [20]).
On the other hand, the results from the most general, complete case
(Model B) differ from the sech2 law at all radii, and the difference is large –
by as much as 30-40 % at R = 18−22 kpc. This may be explained as follows.
In this radial range, the disk density is low, and the vertical distribution is
extended [19]; thus, the radial term in the Poisson equation may not be
negligible. Hence, the inclusion of the R term in the disk Poisson equation,
and the other kinematic terms in the Jeans equations, has a substantial effect
on the density distribution. Recall that in this region, the rotation curve
is taken to be flat. Hence, the difference between the results from these
two models at large radii is purely due to the kinematic terms involving
the velocity dispersion and the cross terms in the radial and vertical Jeans
equations.

86
 #
"
! ' )*+,-(./05
$%& 3
2
1
7
64
8
9 
!

 

#
"&
$
% (
'




)
,
*
+
-
.
4
2
3

0
/
1
5
6


 
Figure 21: Plot of resulting stellar vertical density, ρ(z) vs. z at R = 8.5 kpc and R = 18
kpc, respectively. The solid curve represents sech2 distribution; and the dashed curve
represents the density distribution obtained using the complete, general model (Model B)
– see the text for details. The difference between the two distributions is substantial in
the outer Galaxy (R=18 kpc), while it is small in the solar neighbourhood (R=8.5 kpc).
This shows the necessity of using complete Poisson and Jeans equations to obtain ρ(z) for
stars, in the outer Galaxy or in low density regions. Source: Taken from [20]

At radii less than 7 kpc, the difference between the mid-plane densities in
model B and sech2 case is again quite significant ≤ 20%. This is attributed
to the non-flat rotation curve and the stellar velocity dispersion. At an
intermediate radial range of 7-12 kpc, the change as seen in Model B from
sech2 is small ∼ a few %. By sheer coincidence, the use of sech2 as is routinely
done for the stellar disk in the literature; happens to be valid reasonably well
in the intermediate radial range, including in the solar neighbourhood (R
= 8.5 kpc). The changes from sech2 results shown in this radial range do
not show any clear pattern with radius. It is the complex interplay among
the various kinematical terms present in Eq.(4.18) that sets the value of the
mid-plane density and the difference with the sech2 model results, including
the sign difference.
As an example, in Fig 21, the resulting density distribution for Model B
and sech2 model are shown for R=8.5 kpc and 18 kpc (left and right panels,
respectively). Note that the density distribution for the general, complete

87
model (Model B) shows a substantial difference from the standard sech2
model in the outer Galaxy, at R=18 kpc, while the difference is small at
R=8.5 kpc. This shows the importance of using the complete Poisson and
Jeans equations while solving for ρ(z) for stars, in the outer Galaxy or in
region of low density.
The conclusion from this study is that at larger radii and in regions of low
density or a thick disk; the above complete, general model should be employed
in future studies to get accurate results for the vertical disk distribution. The
predicted changes in the vertical density profile due to various kinematical
effects considered in the model are possible to be checked against the new,
accurate data, e.g. from Gaia DR3 and DR4 or LAMOST.
A few general comments are as follows. First, the resulting density profiles
for the general case (model B) can be fit to a function of type sech2/n , but
then n is fund to vary with ∆z. This trend was also found for the multi-
component thin disk case studied by [19], also see Section 4.2.6. Thus, sech2/n
is not valid for the density profile in this case as well. See Appendix A for a
detailed discussion on the vertical density profiles, and scale heights. Second,
in the general case (model B), the disk shows flaring and the flaring over the
radial range R=2 to 22 kpc is smaller by a factor of 2 (reducing from a factor
of ∼ 16 to 8) compared to that for the standard sech2 law resulting for a
thin, stars-alone disk (see Table 2 in [20]). Thus, flaring of the stellar disk in
the outer disk region appears to be a generic result (also, see Section 5 for
further details).

4.4.2. Model for a non-isothermal galactic disk

The vertical density distribution of stars in a galactic disk is tradition-
ally obtained by assuming an isothermal velocity dispersion of stars, namely
dispersion that is constant with z, for simplicity. Only a few papers have ex-
amined the dynamical consequences of considering a non-isothermal velocity
dispersion [13, 154]. One reason could be that observationally the velocity
dispersion was not known in detail at different heights. Interestingly, [13, 154]
had studied the vertical density distribution for the case when the velocity
dispersion increases with height from the mid-plane, and showed that a stable
solution was possible.
Recently, high-quality, high-resolution data: such as obtained using Sloan
Digital Sky Survey (SDSS); Large sky area multi-object fiber spectroscopic
telescope (LAMOST); Radial Velocity Experiment (RAVE); and Gaia shows
that the vertical velocity dispersion, σz , for stars increases with z, the dis-

88
tance from the mid-plane.14 This is shown for various tracer stars; and
even over different metallicity bins; as seen in the solar neighbourhood, and
also at larger radii (see [21], and references therein). This shows that the
non-isothermal velocity dispersion is a genuine physical feature of the stellar
disc. A few of the above papers discuss the vertical variation of the velocity
dispersion, particularly for the thin disc of stars [155, 156, 100, 102, 157].
Although a few papers, based on the observed data, study the effect of
non-isothermal velocity dispersion on the measurement of dynamical quanti-
ties – such as, the local dark matter estimate or the total mid-plane density
(i.e., the Oort limit), see e.g.,[158, 100]; they do not study the effect of the
non-isothermal dispersion on the vertical stellar density distribution.
Formulation of Equations
[21] study the dynamical effect of such non-isothermal velocity dispersion
on the self-consistent vertical distribution of thin disk stars in the Galaxy
– both for a single-component stellar disk, as well as for stars in the multi-
component disk plus halo system. In both cases, they find that the inclusion
of non-isothermal dispersion has a significant effect on the vertical density
distribution, as summarised next. Although the numerical results are ob-
tained for the input parameters for the Galaxy; the formulation as well as
the trends in results obtained in this work are general, and are applicable for
a typical galactic disk.
[21] note that they model the stellar disk to consist of a single component,
for simplicity. That is, they treat stars of different ages or metallicity values
of the thin disc cumulatively, as is the standard practice in studies of vertical
disk structure (e.g., [22]; also, see discussion in Section 4.2.1).
For the stars-alone disk, when the vertical velocity dispersion, σz , is a
function of z; the joint Poisson-hydrostatic balance equation (Eq.(3.3)) is
modified to:

2
σz2 ∂ 2 ρ σz2 ∂σz2 ∂ 2 σz2
  
∂ρ 1 ∂ρ
− 2 + + = −4πGρ (4.20)
ρ ∂z2 ρ ∂z ρ ∂z ∂z ∂z2

Next, σz is taken to increase linearly with z, based on most of the observed

data in literature, and is given as:

14
In this subsection, σz is taken to denote the vertical velocity dispersion for stars, in
keeping with the notation used in [21], and is given without the subscript s.

89
 σz = σz,0 + Cz (4.21)
where C is the linear gradient (dσz /dz) in velocity dispersion along z and
σz,0 is the dispersion at z=0, i.e, the galactic mid-plane. Substituting this
expression into the joint Poisson-hydrostatic balance equation (Eq.(4.20)), it
gives:

 2
∂ 2ρ −4πGρ2 1 ∂ρ
= +
∂z 2 (σz,0 + Cz)2 ρ ∂z
(4.22)
2C 2 ρ
 
2C ∂ρ
− −
(σz,0 + Cz) ∂z (σz,0 + Cz)2

The solution of this equation gives the self-consistent, non-isothermal ver-

tical density distribution ρ(z) of a stars-alone disc for a linearly increasing
vertical velocity dispersion. It is assumed that such a stable solution exists.
Eq. (4.22) is solved using the same numerical procedure as in [16]. The
corresponding isothermal case is solved numerically by setting the gradient
C = 0. Alternately the solution for the isothermal case is simply given by
the sech2 law (Eq.(3.4)).
In an analogous way, the joint Poisson-hydrostatic balance equation that
governs the vertical distribution of stars for a multi-component disk plus halo
model for the non-isothermal case, is obtained to be:

∂ 2 ρs
 
ρs ∂(Kz )h
= −4πG (ρs + ρHI + ρH2 ) +
∂z 2 (σz,0 + Cz)2 ∂z
 2
1 ∂ρs
+ (4.23)
ρs ∂z
2C 2 ρs
 
2C ∂ρs
− −
(σz,0 + Cz) ∂z (σz,0 + Cz)2

The gas components, HI and H2 gas, both have isothermal dispersions

along z, as supported by observed data. Therefore, the following equation
governs their vertical distribution (as was given by Eq.(4.4) for a thin disk
case, see Section 4.2.1):

90
 ∂ 2 ρi
 
ρi ∂(Kz )h
= 2 −4πG (ρs + ρHI + ρH2 ) +
∂z 2 σz,i ∂z
 2 (4.24)
1 ∂ρi
+
ρi ∂z
for i= HI & H2 respectively. Eq.(4.23) & Eq.(4.24) are solved together
for a coupled system, following the approach as in [16]; to obtain ρi (z) for
i = s, HI, H2 , corresponding to stars, HI and H2 , respectively.
The corresponding case when the stellar dispersion is isothermal, is treated
by setting C = 0 in Eq. (4.23) and this equation is solved together with Eq.
(4.24) to obtain the density distribution for the three disk components in the
coupled case.
Input parameters and results
An important input parameter for this work is the non-isothermal velocity
dispersion. The mid-plane value of the vertical stellar velocity dispersion,
σz,0 , is obtained from observations of [56], see the discussion in [19, 21].
The observed stellar velocity dispersion shows a vertical velocity gradient,
as determined by different surveys: such as, TAGS, RAVE [100]; LAMOST
DR2, and Gaia [102]; and SDSS and LAMOST surveys [155], for the thin
disk stars. The observed velocity profiles of thin disk stars in the solar
neighbourhood are found to be similar irrespective of the tracer or the survey;
hence, the measured linear gradient values are expected to lie in a small
range. [21] note that the observed gradient values thus obtained cover thin
disk populations with a finite range of metallicities and ages; thus, the net
average velocity gradient value is applicable to the whole thin disk treated
as a single component in their work.
In view of these points, [21] adopt the value for the gradient along z
for the stellar velocity dispersion to be +6.7 km s−1 kpc−1 as determined
from observations for the radial range of 6-10 kpc [100] as the standard
value to explore the effect of the non-isothermal velocity dispersion. They
also consider a higher value of +10 km s−1 kpc−1 later for a few cases to
illustrate the quantitative variation with the gradient, and state that such
higher values are likely to occur at large radii due to external tidal encounters.
The other observed input parameters for the Galaxy are as given by [19,
21]. An interesting point is that during the integration, σz , the z velocity
dispersion of stars is kept constant at its value at z =1.5 kpc (the last observed

91
 5 ×10 2.5 ×10
2 2

stars_alone_iso stars_alone_iso
stars_alone_grad6.7 stars_alone_grad6.7
4 stars_alone_grad10 2.0 stars_alone_grad10
(z) (M pc 3)

(z) (M pc 3)
R=8.5 kpc R=10 kpc
3 (b) 1.5 (c)

2 1.0

1 0.5

00 250 500 750 1000 1250 1500 1750 2000 0.00 250 500 750 1000 1250 1500 1750 2000
Vertical Distance z (pc) Vertical Distance z (pc)

Figure 22: Plot of resulting self-consistent vertical density distribution, ρ(z) versus z for
a stars-alone disk, for the isothermal case (shown as a solid curve); and for the non-
isothermal cases for a vertical velocity gradient of 6.7 kms−1 kpc−1 (the observed value)
and 10 kms−1 kpc−1 (shown as dashed and dashed-dot curves, respectively). The results
for R=8.5 kpc and 10 kpc are shown in the left and right panels, respectively. The higher
vertical pressure in the non-isothermal case results in a lower mid-plane density,ρ0 , and
a higher scale height (HWHM) compared to the isothermal case. This gives rise to a
distribution that is more extended along the vertical direction, or, it is flatter. This effect
is stronger for a higher velocity gradient (due to higher pressure), and also at larger radii
(due to lower disk self-gravity). Source: Taken from [21]

point) and beyond for all higher z points. For comparison, the corresponding
isothermal cases are also solved, as described above.
Results
The resulting vertical stellar density distribution vs. z for the stars-alone
case is plotted for the isothermal case and the non-isothermal case, at R=
8.5 kpc, for a vertical velocity gradient of 6.7 (the observed value) and 10
km s−1 kpc −1 , respectively (see Fig. 22, left panel). A similar plot for R=
10 kpc is given in the right panel of Fig. 22.
At each radius, the non-isothermal density distribution, ρ(z), is found to
have a lower mid-plane density value, and a higher scale height (HWHM)
value, than for the corresponding isothermal ρ(z) distribution; and is thus
more extended along the vertical direction, or, is flatter. For example, at

92
R=8.5 kpc, the mid-plane density is lower by 32.6 percent, and the HWHM
is higher by 37.1 percent, than the isothermal case for the observed gradient
of 6.7 km s−1 kpc−1 .
This effect is stronger for a higher velocity gradient at a given radius –
compare the dashed and the dashed-dot curves in each of the two panels in
Fig. 22. The density distribution for a gradient of 10 km s−1 kpc−1 is flatter
than for the lower gradient value; so that the mid-plane density at R=8.5
kpc is lower by 41.9 percent and the HWHM value is higher by 53.7 percent
than the isothermal case, compared to the smaller corresponding changes for
the smaller velocity gradient as given above.
The physical reason for the extended distribution is that in the non-
isothermal case with a positive velocity gradient, the vertical pressure is
higher than the isothermal case at each z. Therefore the hydrostatic balance
between the self-gravity and the non-isothermal vertical pressure gradient
results in a density distribution that is vertically more extended than the
isothermal case. This effect is stronger as z increases, since the dispersion
is higher, hence the density distribution becomes flatter with increasing z.
Obviously, this effect is stronger for a higher value of the velocity gradient.
Further, for a given velocity gradient, this effect is more prominent at
larger radii. This is because the self-gravity of the stellar disk decreases with
increasing radii hence the non-isothermal velocity dispersion can affect the
vertical density distribution more (compare the two panels in Fig. 22). Since
the HWHM is higher for the non-isothermal case, and its value increases for
a higher radius; the stellar disk shows flaring, which is higher than seen in
the isothermal case. Further, this flaring increases with the velocity gradient
(see Fig. 2 in [21]).
The density profile for the single-component non-isothermal case thus
shows a deviation from the standard sech2 profile; and this deviation increases
at high z. Similar deviation is seen at high z in the observed intensity
profiles, and is known as the ”wing”. This will be discussed more later in
this subsection.
Next, consider the effect of non-isothermal velocity dispersion of stars in
a more realistic case, namely on stars in a multi-component disk plus halo
system. Here too, on inclusion of the non-isothermal velocity dispersion,
the results show the same trend; but the effect of non-isothermal dispersion
is reduced due to the opposite, constraining effect of gas and dark matter
halo gravity. Fig. 23 shows the resulting density distribution of stars in
a multi-component disk plus halo case vs. z, at R=8.5 kpc, for the non-

93
 ×10 2
6 stars_coupled_iso
stars_coupled_grad6.7
5 stars_alone_grad6.7

(z) (M pc 3)
4 R=8.5 kpc
(a)
3
2
1
00 250 500 750 1000 1250 1500 1750 2000
Vertical Distance z (pc)
Figure 23: The non-isothermal density vertical density distribution of stars versus z
is shown at R=8.5 kpc: for the stars-alone case (dashed-dotted curve); for the multi-
component disk plus halo case (dashed curve)- both the above obtained using the disper-
sion gradient of +6.7 km s−1 kpc−1 ; and the stellar distribution versus z for the isothermal
case for the multi-component disk plus halo model (solid curve). The non-isothermal stel-
lar distribution in the coupled case is more constrained towards the mid-plane, due to the
gravity of gas and dark matter halo; compared to the non-isothermal stars-alone case. But
it still remains more extended than the isothermal stellar distribution in the coupled case.
Source: Taken from [21]

isothermal case with the observed velocity dispersion gradient of 6.7 km s−1
kpc−1 . Also shown are results for stars in a coupled multi-component disk
plus halo case with isothermal velocity dispersion, and for a single-component
stars-alone, non-isothermal case. Interestingly, although the gas and dark
matter halo constrain the non-isothermal stellar vertical distribution in the
multi-component case (compared to the non-isothermal, stars-alone case),
the distribution remains more extended that the isothermal stellar solution
of the coupled case. Thus the effect of non-isothermal dispersion is opposite
to the constraining effect of the gas and halo gravity, and dominates over it.

[21] show that the reduced mid-plane density of the stellar distribution
due to the effect of non-isothermal velocity dispersion, will be reflected in a
lower value of the theoretical estimate of the total mid-plane density, or the
Oort limit, by ∼ 15% compared to a similar theoretical estimate obtained in

94
 3.0 ×10 3.0 ×10
2 2

stars_alone_grad6.7 stars_alone_grad6.7
best-fit 0sech2(z/z0) 0,thinsech2(z/zthin)
2.5 2.5 0,thicksech2(z/zthick)
R=8.5 kpc best-fit 2
double sech
2.0 2.0
(z) (M pc 3)

(z)(M pc 3)
(a)
R=8.5 kpc
1.5 1.5
(b)
1.0 1.0
0.5 0.5
0.00 250 500 750 1000 1250 1500 1750 2000 0.00 250 500 750 1000 1250 1500 1750 2000
Vertical Distance z (pc) Vertical Distance z (pc)

Figure 24: The resulting non-isothermal vertical stellar distribution obtained for a velocity
gradient of 6.7 km s−1 kpc−1 for stars-alone case at R= 8.5 kpc: as fitted by a single and
double sech2 profiles (left and right panels), respectively. The single sech2 function (left
panel) gives a poor fit to the density distribution at all z values, especially at the ”wing”
or high z part of the distribution. The right panel shows the best-fitting double sech2
profiles consisting of ”a thin and a thick disk” (shown as dashed-dot and dotted curves,
respectively) which together give a good fit to the density distribution of a non-isothermal,
stars-alone case (solid curve) at all z. This shows how an observer might misinterpret a
non-isothermal single-component stellar disk density distribution as a superposition of two
separate sech2 or isothermal disks. Source: Taken from [21]

the isothermal case [19].

Observed ”wings” at high z due to non-isothermal velocity disper-
sion
The vertical profile of an external galaxy is often found to deviate from
the single sech2 function (Section 2), especially at high z. The excess flux
at high z makes the observed distribution look like a wing w.r.t. the fitted
sech2 function. To account for this wing, another sech2 profile representing
a thicker disk is typically added during the fitting to the observed data to
obtain the parameters of the two disks (e.g.,[159]).
[21] note that the non-isothermal ρ(z) distribution for a single stellar disk
shows a similar deviation as seen in the observed profiles; and this could well
be the cause of the observed ”wings” in the intensity profile at high z. To

95
check this idea, they fit the model non-isothermal ρ(z) distribution obtained
at R= 8.5 kpc using the observed velocity gradient value of 6.7 km s−1 kpc−1
– first, by a single, and then by double sech2 profiles, see Fig. 24.
It is clearly seen that even the best-fit single sech2 function gives a poor
fit to the model ρ(z) distribution. On the other hand, the double sech2 profile
fits the model results for the density distribution very well at all z (see Fig.
24). Hence [21] caution that if an observed luminosity profile shows a wing at
high z and a single sech2 function is found to give a poor fit to the data, this
may well be due to a non-isothermal origin of the ρ(z) distribution of stars
of a single stellar disk. Thus, evoking a second, thicker disk – as is routinely
done in the literature – to fit the observed profile that shows a wing at high
z, could be redundant. A similar conclusion also holds for a non-isothermal
stellar disk in the coupled case (see Fig. 5 in [21]).
Further, even when a genuine distinct Thick disk is present – as evident
from its distinct chemical and kinematical properties – see Section 1; a part
of the deviation from the sech2 distribution could be due to a non-isothermal
velocity dispersion in the disk. Hence, [21] stress that this should be taken
into account while obtaining the properties of the two disks by the usual
double sech2 fitting procedure.
An interesting point is that at large R, such observed wings or a broader
distribution at high z, could also arise in the isothermal case for the multi-
component disk plus halo model; as was shown for UGC 7321 ([91]; see
Section 4.3.2). In that case too, it was argued that evoking a second disk to
explain the wings could be redundant.
Limit of validity of the isothermal assumption:
To determine the limit of validity of the isothermal assumption normally
assumed, [21] try varying the values of the gradient, and show that even
a small gradient of +2 or +3 km s−1 kpc−1 is found to cause a change of
-9.3 % and -14% in the mid-plane density value of the stellar distribution
compared to the isothermal case, for the stars-alone case. The corresponding
changes are somewhat smaller, namely -8.3% and -11.7% due to the opposite,
constraining effect of gas and halo in the the multi-component disk plus halo
case compared to the corresponding isothermal, stars-alone case. Such small
gradients are likely to occur in real galaxies, at least locally. In fact, a strictly
constant – or isothermal – velocity dispersion that is typically assumed, may
be an idealized case. Hence, [21] alert that the isothermal assumption, or the
resulting sech2 profile, that is routinely used to describe the stellar vertical

96
distribution in the literature, may not be strictly valid even if there is a small
velocity gradient of ∼ 3 km−1 kpc−1 . This warning is of practical importance
since the resulting small changes in mid-plane density are now possible to be
detected by means of accurate measurements of star counts data, e.g. using
Gaia.
The above results show that the observed non-isothermal stellar vertical
velocity dispersion has an important dynamical effect on the vertical disk
structure. That is, it results in a self-consistent density distribution that
is substantially more extended along the vertical direction, or it is flatter
compared to the isothermal case. Hence this effect should be included in the
future dynamical modelling of a galactic disk.

4.5. Applications to observational data and theoretical studies in the litera-

ture
The multi-component disk plus halo model developed above, for the thin
disk and the thick disk cases (Section 4.1 to Section 4.4), has been exten-
sively applied in the literature to theoretical studies, and also to analyze
observational data. We highlight a few examples here. Because of the phys-
ical rigour of the model, and the complete treatment that involves the disk
components (stars and gas) and the halo; this model has been preferred in
accurate studies of disk vertical structure – over the usual sech2 model – even
though this involves more work.

4.5.1. Applications to analyze intensity profiles

In an important paper, Comeron et al (2011) [160] applied the model
by [16] to analyze the 3.6 µ images of a sample of 46 edge-on galaxies from
the S4 G survey [27]. The aim was to constrain the thin and the Thick disk
properties of external galaxies. Here a Thick disk is kinematically and chem-
ically distinct from a thin disk. [160] point out that it is necessary to follow
this rigorous model which gives physically motivated density profiles that
are more accurate. Next, they match the corresponding model luminosity
profiles with observations. They follow this procedure rather than taking the
simpler approach of decomposing the observed stellar profiles in terms of a
linear superposition of two sech2 disk components, each with a scale height
that remains constant with radius. The latter approach is usually used in
the literature because the trial density functions then can be taken as simple
analytical functions. [160] multiply their model density profiles with trial

97
values of mass-to-light ratios for the thin and thick disks respectively, and
match the total with the observed luminosity profiles from the S4 G data, so
as to obtain the disk parameters for both the disks. The best-fit to the data
indicated that the thick disks were more massive than was generally believed
till then, and could be comparable in mass to the thin disk. From this they
proposed a galaxy formation scenario where the thick disk is formed first.
[160] stress that an accurate determination of the relative masses of thin
and Thick disks – which was made possible because of the use of physically
meaningful density profiles obtained – allowed them to identify the formation
mechanism for the thick disk. The formation mechanism of the thick disk
had heretofore been a challenging problem.
It should be noted that the halo contribution is not included in this work
for the sake of simplicity, and because the uncertainties in the mass-to-light
ratio values used are stated to be comparable to the effect of neglect of the
dark matter halo in the problem [160].
In another, earlier application of the model, [35] considered real data
(from [22]) for intensity profiles in two external galaxies. By analyzing these
correctly (See Section 2.2 and Section 4.2.7), [35] showed that the disk scale
height increases by a factor of 2-3 within the optical disk. Further, by apply-
ing the multi-component disk plus halo model, [35] showed that this flaring
in the disk constraints the parameter Rv , to lie in the range of 2 − 3 RD .

4.5.2. Applications to the study of gas distribution

Although this review has mainly focused on the stellar disk; the multi-
component disk plus halo model simultaneously gives the resulting self-consistent
vertical density distribution including the disk thickness for each disk compo-
nent, including gas (Section 4.2.2). Getting accurate gas scale height values
is important for a number of physical processes, such as star formation. We
give a few examples where this model has been applied so as to obtain the
accurate vertical density distribution of gas and the scale height:
[161] has applied the above model, and employed a new, iterative ap-
proach, to simultaneously obtain the best-fit values of the HI velocity disper-
sion and the gas thickness in a sample of spiral galaxies, by making use of the
3-D HI data. First, the model results for the vertical density profiles of HI
and H2 gas as a function of radius are obtained, using the multi-component
disk plus halo model in the thick disk case (Section 4.3). To do this, the
observed values of the stellar and gas surface densities in the literature as
well as the H2 velocity dispersion are used as input parameters for the model.

98
Then, by fitting the model results with the 3-D HI data cube in an iterative
fashion, the best-fit HI velocity dispersion and the best-fit HI scale height as
a function of radius are obtained simultaneously. The best-fit HI dispersion
shows a variation with radius (see [161] for details).
In another application, a self-consistent, multi-component disk analysis
was done to obtain the vertical distribution and hence the disk thickness
(HWHM) of the HI gas, as a function of radius in the ultra-diffuse galaxy
AGC 242019 [162]. The formation mechanism for ultra-diffuse galaxies are
not yet well-understood. [162] found the HI scale height values in AGC
242019 to be moderate, and comparable to that for a sample of dwarf galaxies.
Hence they conclude that stellar feedback has not been important in the
formation of this ultra-diffuse galaxy.

4.5.3. Applications to theoretical studies

The model has been applied to a number of theoretical studies in the
literature to get a more accurate description of the vertical disk structure
and stability; star formation; as well as to understand the halo properties, as
discussed next. The change in the density profile including the thickness in
the multi-component case, compared to the standard sech2 solution for the
one-component case; and the associated change in the gravitational poten-
tial, can affect various dynamical processes in the disk. Some examples of
application of this model for theoretical studies are:
In an interesting application, [163] use the model to develop the theory
of stability of 3-D gas disk in a galaxy. To do this, [163] use the multi-
component disk plus halo model in the thin disk case ([16]; also, see Section
4.2.2 here) to set up a gas disk in hydrostatic equilibrium. This allows them
to treat the problem locally at a given radius, R (see their Section 4.2);
which simplifies the set-up for numerical simulations as no radial exchange of
information is needed. Their analysis highlights this subtle point. Next, [163]
show that using this initialization condition for their numerical simulations
allows them to study the onset of gravitational instabilities in a 3-D gas disk
more accurately. This has implications for resulting star formation rates, and
formation of spiral arms.
In another study, the multi-component disk plus halo model was applied
[130] to data for a set of spiral galaxies to obtain the redistributed or enhanced
mid-plane density. Assuming the star formation to depend on a power, n′ ,
of mid-plane density, and comparing with the observed star formation rate,
they find the best-fit value of n′ = 1.5. Hence [130] conclude that the Schmidt

99
law for star formation rate (∝ ρn , where n = 2), as given by [164], is not
universal.
[165] point out that an accurate stellar scale height as obtained by the
multi-component disk plus halo model would affect the net pressure on gas
and hence the resulting star formation. Hence they advise using this model
rather than assuming a constant scale height and the ad hoc value of the
stellar scale height taken to be equal to 0.27 RD /5, as is routinely used, for
simplicity, in the literature.
[166] have obtained the vertical distribution assuming multi-component
disk plus halo model, and taking account of the non-isothermal stellar ve-
locity dispersion as in [21]; and have applied these results in the numerical
simulations study of spiral structure in galaxies.
In another application, the multi-component disk plus halo model in the
thin disk case has been applied to study the superthin, LSB galaxies. This
indicates a dense, compact halo in these galaxies that vertically constrains
the stellar disk at all radii. This explains their superthin nature in a generic
way [147].
In an interesting, new approach, the observed HI scale height and rotation
curve values are compared with the model results; to constrain the dark
matter halo parameters such as the density profile and shape of the dark
matter halo in a number of galaxies (see Section 6 for details).

4.5.4. Comments on simplified, non-self-consistent prescriptions for stellar

and HI scale heights in the literature
We warn that in the literature, a variety of simplifying, ad hoc assump-
tions are routinely made, for the sake of convenience, to obtain the vertical
scale heights for stars and gas in a galactic disk; instead of using a rigorous
treatment to obtain a self-consistent density distribution for coupled stars
and gas in a galactic disk as presented in this review (see Sections 4.2-4.4 for
the rigorous treatment).
We point out that the resulting values of scale heights obtained from
the simplified models are not accurate; and when these are used in other,
subsequent analysis, say to study star formation, this could lead to erroneous
results. We mention this point to be comprehensive and illustrate it by
two examples here. Also, see Section 6.4 for more examples of a non-self-
consistent treatment.
Gas disk scale height and star formation

100
 The gas scale height, apart from being of interest for its own sake, is
also of importance for other physical processes such as star formation, and
containment of HI holes. Hence, there is a general interest in obtaining the
gas scale height as a function of radius, and in different types of galaxies.
For this reason, there are several papers where the scale heights for gas have
been estimated, but in a non-self-consistent fashion.
For example, often the gas gravity is neglected and the gas response to the
stellar potential (see Section 3.2, Eq.(3.8)) is taken to determine the gas scale
height. Even when the gas gravity is included [120, 34, 167], the treatment
is not self-consistent. For example, consider a recent detailed work [167]
which does include many physical points, such as gas gravity and the effect
of finite thickness of gas on the gas potential. However, their formulation of
the problem is not rigorously correct, and the resulting density profile is not
self-consistent. Some of the simplifying assumptions in their work are given
next. [167] assume a sech2 stellar density profile; and further assume ad hoc
values of the model parameters (z0 = RD /5) to write the profile. The stellar
potential is taken to be external and constant (that is, not affected by gas or
halo). Thus their formulation of the problem is not self-consistent, and this
will affect the resulting gas scale heights obtained by them.
A correct treatment would involve taking account of the gas and halo.
The resulting self-consistent stellar profile would be steeper than sech2 due
to the constraining effect of gas and halo (see [16, 19]; or, Section 4.2.5),
and hence would result in smaller (true) gas scale height values. Since this
effect is not included by [167], their resulting HWHM values for gas would
be higher than the true values. We caution that this would then result in
smaller predicted star formation rate in their work. This discrepancy will be
seen more at larger radii where the effect of halo on stellar profile becomes
more important (Section 4.2.5; also, [19]). This error will be carried over
in subsequent papers where the non-self-consistent model by [167] has been
applied.
Stellar scale height:
Similarly, in many studies in the literature, instead of treating the proper
coupled system in a self-consistent way; often a number of simplifying as-
sumptions are made so as to get a simple, analytical expression for the stellar
scale height as modified due to the presence of gas. However, as we show
next, these simplifications do not give accurate results and do not represent
a realistic disk.

101
 To give an example, a number of authors (e.g., [168, 169, 170]) use the
following relation between the stellar scale height, hz (in their notation), and
the stellar vertical dispersion,σs , and the surface densities of stars and gas,
Σs and Σg respectively:

hz = σs2 /[2πG(Σs + Σg )] (4.25)

This expression was obtained [169] for a razor-thin gas disk embedded in
a stellar disk. It was obtained by them as a solution to the problem 4.22 in
[171].
This calculation makes a number of serious, simplifying assumptions; and
does not constitute a self-consistent treatment, as discussed next. First,
[169] do not take the gas to be in a hydrostatic balance. They do not take
account of the effect of stars on the gas, as was done in the true coupled
case ([16, 19]; also, see Section 4.2.4, in particular Fig. 6 and Fig. 7). Thus,
the treatment by [169] has not treated the gas component correctly. Second,
the above simplified expression is obtained assuming a sech2 distribution for
stars, hence it does not reflect the steeper stellar distribution profile, or the
corresponding smaller scale height, that is obtained in the coupled case in
the rigorous, self-consistent calculation (see Section 4.2.6; also Appendix A).
Thus, the above expression is not valid for a true coupled system. The
stellar scale height values for the Milky Way that result from it will differ
from the results obtained in the self-consistent calculation by [16, 19]. The
latter results agree well with observed values, as shown by [16].
Therefore, the use of this simplified, analytical relation Eq. (4.25) as
done in the literature, e.g. by [169] and other papers which cite this work, is
not physically justified. This treatment gives incorrect result for the density
profile of stars, and does not treat the z distribution of gas at all. Thus,
using Eq. (4.25) in subsequent papers will give incorrect results.
Moreover, the application of this result from [169] to a Thick disk and
a thin disk as done by [170] is questionable, since the thin disk has a finite
height (which is just a factor of 2 smaller than the Thick disk scale height -
see Section 7.1). Hence, the thin disk cannot be treated as being razor-thin.
Another serious problem is that the effect of the dark matter halo is
not included in Eq. (4.25). [168] claim that this is justified because the
halo affects scale height by only 12 % in the inner Galaxy, as shown by [16].
However, in the outer Galaxy, the dark matter halo dominates the dynamics;
and the stellar scale height decreases by a factor of 2.3 and 3.3 at R= 18 and

102
20 kpc, respectively (see Table 2 in [19]; also, see Fig. 11, panel for R=18
kpc here). Thus ignoring the effect of halo on the stellar disk as done in the
above papers [168, 169, 170] is a serious error; and this will give spuriously
large stellar scale height values compared to the ”true” stellar scale height
values. The latter are as given by a self-consistent treatment, as in [16, 19].
We caution again that the above expression (Eq. (4.25)) is incorrect, and
adopting it in subsequent papers would affect the accuracy of the results
obtained.

4.6. Gravitational potential energy of a multi-component galactic disk

A generic result from the multi-component disk plus model is that the
vertical disk distribution is constrained closer to the mid-plane (See [16, 19];
or, Section 4.2.5). To understand the implications of this result for the
energetics of the disk, the gravitational potential energy of a multi-component
disk was studied by Sarkar & Jog (2022) [90].
The gravitational potential energy of spherical and spheroidal/ellipsoidal
gravitating objects has been extensively studied in the literature [172, 2]. In
contrast, the potential energy of a thin disk system has not received much at-
tention, even though it is of direct relevance for realistic systems, namely the
disk galaxies. One reason for this is that the geometry being ”open” in this
case makes it a harder problem, than say, the spheroidal case that is usually
considered. This problem was studied by Camm (1967)[173] who considered
an isothermal, one-component, self-gravitating thin disk of constant surface
density and infinite extent in the plane. By using analytical techniques and
potential theory, [173] obtained the gravitational potential energy per unit
area of such a stars-alone disk (see Eq.(4.26)), as briefly discussed below.
Normally the potential energy of a mass distribution in a finite volume
is defined as the energy released in assembling the finite system from an
infinitely dispersed state, and it is finite. For a thin disk distribution, the
state of collapse of the disk mass on the z = 0 plane is defined to be the
state of zero potential energy. The work done that is required to build the
disk of finite height, from that state, is considered to be the potential energy
stored in the disk, so that it has a finite value – for a detailed discussion see
[173, 90]. Therefore, the potential energy of the disk is defined in terms of
the energy contained in a column of unit cross section, that is, as potential
energy per unit area of the disk.
The mathematical expression for the gravitational potential energy of
a thin, single-component, isothermal disk of constant surface density was

103
derived by [173] and is given in terms of the integral over ρ(z), the mass
density; Kz , the force per unit mass due to the self-gravity of the disk; and
z; and is equal to15 :
Z +∞ Z +∞
∂Φ
W= ρ(z) z dz = − ρ(z)Kz z dz (4.26)
−∞ ∂z −∞

Using the symmetry of ρ(z) w.r.t. z, this can be written as:

Z +∞
∂Φ
W=2 ρ(z) zdz (4.27)
0 ∂z
4.6.1. Motivation and Formulation
We stress that a real galactic disk is a multi-component system of gravita-
tionally coupled stars and gas embedded in the potential of the dark matter
halo ([16, 19]; also, see Section 4.1-Section 4.3). [90] follow the method devel-
oped by [173] for a self-gravitating one-component disk; and systematically
extend it to a gravitationally coupled, multi-component disk, and thus ex-
plicitly derive the expression for the potential energy per unit area for the
multi-component case. We emphasize that this is not a trivial extension, see
[90] for the detailed steps of the extensive calculation. For simplicity, the
dark matter halo is not included, hence their calculation is mainly applicable
to the inner Galaxy. The main points regarding the calculation are sum-
marized next. It should be pointed out that this calculation as well as the
results from this section are valid for any general galactic disk, although the
numerical results are obtained for the Milky Way since the input parameters
are known for it.
While deriving Eq.(4.26), [173] had used only the z component of the Pois-
son equation, since an infinite disk of constant surface density was treated.
[90] show that this assumption is also justified for a general, thin galactic
disk (for which the z term of the Poisson equation is >> the R term - see
[2]; also, Section 3.1 of this review). Hence the vertical density distribution,
force and energy become only z-dependent quantities, and depend on the
local value of the disk surface density, Σ. Hence the calculation is local, and
the potential energy obtained at a given radius is independent of the value
of the surface density at other radii.

15
In writing this result from [173], the standard notation for the Poisson equation is
used, which is ∂ 2 Φ/∂z2 = 4πGρ. See Section 3.1, and [90] for details.

104
 Therefore, [90] could follow the same approach as in [173], while deriving
the potential energy for a realistic, multi-component galactic disk; with stellar
and gas disks with surface densities that vary with radius. The potential
energy per unit area for the coupled case is then obtained to be:
Z ∞
∂Φcoupled
Wcoupled = z (ρs + ρg ) dz (4.28)
−∞ ∂z
Thus, analogous to the one-component case, the potential energy for the two-
component case is obtained to be an integral over the total mass density, the
total gravitational force, and the vertical distance z.
The above equation can be separated into two terms; and using the sym-
metry of ρ(z) about z = 0, these can be further simplified as:

Z ∞ Z ∞
Wcoupled = −2 z Kz,coupled ρs dz − 2 z Kz,coupled ρg dz (4.29)
0 0

where kz,coupled is the joint gravitational force per unit mass in the coupled
system. The integrals in this equation can be considered to represent the po-
tential energy per unit area of the stellar disk and the gas disk, respectively,
in the coupled stars plus gas system. This separable form results because the
two disk components are now being built from z=0 to their corresponding
vertical distributions simultaneously, against the same coupled force. Note
that in the limit of ρg → 0, or ρs → 0, Eq.(4.29) goes over to the correspond-
ing stars-alone and gas-alone cases, respectively.
The above formulation can be extended in a similar fashion to any n-
component disk system (for n > 2). This has been done for a three-component
disk consisting of stars and two gas components (HI and H2 ) in the Milky
Way (see [90] for details). Also, the above formulation could be extended to
a multi-component stellar disk (n > 3) for the many stellar sub-components
that have now been identified from Gaia. The study of the vertical distribu-
tion of the latter system has been mentioned as a possible future problem in
Section 8.

4.6.2. Potential energy for a disk in hydrostatic equilibrium

For a disk in a vertical hydrostatic equilibrium, ρ(z) and Kz used in
the above expression for the potential energy, are related to each other by
the equation of hydrostatic equilibrium. The self-consistent solution for the

105
 2.00
stars_alone
1.75 gas_alone
force on stars or gas
in stars+gas coupled disc
1.50
|kz|(km 2s 2pc 1)
1.25
1.00 R=8.5 kpc
0.75
0.50
0.25
0.000 250 500 750 1000 1250 1500 1750 2000
Vertical Distance z (pc)
Figure 25: Plot of the vertical force per unit mass, |Kz |, versus z at R=8.5 kpc, for stars-
alone and gas-alone under their own self-gravity (denoted by solid and dashed curves,
respectively). The force due to gravitationally coupled stars and gas – felt by both stars
and gas (shown by the dashed-dotted curve) is larger than the individual one-component,
gravitational force for stars and gas, at each z. Source: Taken from [90]

vertical density distribution, ρ(z), and its derivative have to be obtained nu-
merically by solving the coupled, joint Poisson-hydrostatic balance equation
(see Eq.(16) from [90]). The joint force, Kz,coupled in the coupled case at each
z can be obtained from the equation of hydrostatic balance for any one disk
component in the coupled case (see Eq. 14 from [90]), in terms of the values
of ρ and dρ/dz for that disk component obtained numerically, as described
above. Similarly, for the one-component case, the values of Kz and ρ at any
z are obtained using the analytical solutions (see Eq.(6) and the discussion
following that in [90]).
Using the above approach, and the observed input parameters for the
Milky Way, the values of the force Kz for stars and gas respectively and for
the coupled force Kz,coupled are obtained, and their magnitudes are plotted in
Fig.25. At each z, the joint gravitational force |Kz,coupled | is larger than the
one-component values.16

16
Note that, we have seen the effect of this larger force in the coupled case in constraining
the disk components in the earlier sections; but the fact that the magnitude of the force

106
 1.0
10 stars_alone gas_alone
stars in stars+gas gas in stars+gas
z (z) kz (M pc 3km2s 2)

z (z) kz (M pc 3km2s 2)
coupled disc 0.8 coupled disc
8 R=8.5 kpc R=8.5 kpc
(a) (b)
0.6
6

4 0.4

2 0.2

00 500 1000 1500 2000 2500 3000 3500 4000 0.00 500 1000 1500 2000 2500 3000 3500 4000
Vertical Distance z (pc) Vertical Distance z (pc)

Figure 26: Plot of the integrand −zρ(z)Kz of the gravitational potential energy per unit
area versus z for stars and gas (HI) at R= 8.5 kpc; obtained for a one-component case and
then for the coupled case. The left panel shows the plot of energy integrand for stars-alone
(solid curve) and in the coupled case (dashed curve). Twice the area under the curves
gives the gravitational potential energy per unit area in the two cases; which, surprisingly,
is obtained to be the same. The right panel shows a similar result for gas. Thus, the net
energy per unit area for each of the two disk components remains unchanged. See the text
for the physical explanation for this surprising result. Source: Taken from [90]

Next, the corresponding integrands of the potential energy for the one-
component case (solid line) and the same component in the coupled case
(dashed line), obtained using Eq.(4.27) and Eq.(4.29) respectively, are plotted
in Fig. 26 (panel a for stars and panel b for gas). In each case, twice the
area under the curve gives the potential energy per unit area.
Interestingly, despite being in the higher gravitational force in the coupled
system, the values of the gravitational potential energy per unit area for both
stars and gas turn out to be the same as their corresponding values for the
single-component, self-gravitating cases. This is because in the coupled case,
while |Kz | is higher, the density distribution is constrained to be closer to the
mid-plane, hence the integrand peaks at a smaller z. Therefore the energy

is higher in the coupled case was not explicitly shown in any of the papers discussed in
Section 4.1 to Section 4.5.

107
integrand is now redistributed along z while conserving the total energy.
Thus, the work done required to build up a stars plus gas disk turns out to
be the same as the sum of energies required to build separate one-component,
self-gravitating stellar and gas disks of same parameter values.
Analytical simplification for potential energy per unit area of a
disk:
To investigate the physical reason for this surprising result, the expres-
sion for the potential energy above is simplified analytically by writing Kz in
terms of the equation of hydrostatic balance. On substituting Kz,coupled ob-
tained in terms of the equation of hydrostatic balance for any one of the disk
components in the coupled case (see Eq. (14) from [90]) in the expression for
the net potential energy per unit area in the coupled case (Eq.(4.29)), the
latter reduces to:

2 2
Wcoupled = σz,s Σs + σz,g Σg (4.30)
Here the notation as in [90] has been used to denote σz,s and σz,g as the
velocity dispersion along z for stars and gas, respectively. The expression for
the energy, Wcoupled (Eq.(4.30)), only depends on the intrinsic parameters
for each component; namely, the surface density and the velocity dispersion.
This is equal to the sum of potential energy per unit area for the two com-
ponents taken individually (see [90] for details). This can be understood
physically as follows. Due to the higher gravitational force in the coupled
case (Fig. 25), the self-consistent vertical distribution in each component is
now constrained closer to the mid-plane (or, more of the mass is at a smaller
height), hence the integrand in the expression for the potential energy (Fig.
26) now peaks at a smaller z value, so as to conserve the net potential energy
per unit area. The joint gravity works like an internal force within the sys-
tem, and therefore it can just redistribute the energy (along z) within each of
the two components without changing the total value of the potential energy
in each disk component.

4.6.3. Gravitational potential of a coupled disk

Next, [90] consider another important physical quantity, namely, the work
done or the energy required to raise a unit mass from the mid-plane to a
certain height h in a single-component disk, Ez , i (i= stars or gas); and in
the coupled, two-component (stars plus gas) disk, Ez,i,coupled (i=stars or gas).
The work has to be done against the self-gravity of the stars or gas (in the

108
 1.0 ×10
3

stars_alone
gas-alone
0.8 stars or gas in
stars+gas coupled disc
Ez (km2s 2)
0.6 R=8.5 kpc

0.4

0.2

0.00 200 400 600 800 1000

Vertical Distance z (pc)
Figure 27: Plot of work done, Ez , to raise a unit test mass from the mid-plane to a
vertical height z versus z, at R=8.5 kpc: for the stars-alone case against its self-gravity
(solid curve); for gas-alone case against its own self-gravity (dashed line); and for stars
or gas in the stars plus gas coupled system, against their coupled gravity (dash-dotted
curve). The work required to be done to raise a unit mass to any vertical height is higher
in the coupled case, than in the corresponding one-component cases. This is due to the
higher force in the coupled case (see Fig. 25). This shows that the stars or gas in the
coupled system are more strongly bound to the mid-plane than the corresponding single
component cases. Source: Taken from [90]

single component case); and the joint gravity of stars and gas (in the coupled
case), respectively. Note that this is the measure of gravitational potential
at any height in these cases [89, 93].
For a single-component disk (i = s or g), this work done is given as:
Z h
Ez,i = − Kz,i dz (4.31)
0
where Kz,i represents the self-gravity of the disk.
For the coupled case, it is given by
Z h
Ez,i,coupled = − Kz,coupled dz (4.32)
0
where i= stars or gas. In the coupled case, both stars and gas experience
the same coupled force; hence, the same work has to be done to raise a unit

109
mass of stars or gas to a certain height. Note that in each case, the energy
is positive. Thus, work has to be done to raise mass to a given height.
Using the input values for the Milky way, and obtaining the Kz values as
discussed above (see Section 4.6.2), the work done as a function of z for the
above cases is calculated, for the results see Fig. 27. This shows that to raise
a unit mass to a certain height, higher amount of work has to be done in the
coupled case than in the corresponding single-component cases. This is due
to the higher gravitational force at each z in the coupled case (Fig. 25).
To summarize, while the vertical constraining of a disk component in a
coupled case does not indicate higher gravitational potential energy, it does
mean that each disk component in the coupled case is more tightly bound to
the mid-plane. Hence, each disk component is better able to resist distortion;
say, due to a given external, tidal encounter. A detailed N-body simulation
needs to be done to confirm these predictions.

5. Flaring stellar disk in a galaxy: A generic result

At various points in the previous sections, we have reported that the
scale height of the vertical stellar distribution in a galactic disk increases
with radius. The increase is moderate, a factor of ∼ 2-3, within the first few
radial disk scale lengths, and then increases rapidly beyond that. Thus the
disk thickness or scale height is not constant; rather, the stellar disk flares
with radius. There is clear and convincing evidence for a flaring stellar disk
in observations; as well as in theoretical studies for a typical galactic disk in
hydrostatic equilibrium, when observed input parameters are used. In this
section, we discuss and highlight this striking result and show that a flaring
stellar disk is a generic feature, and discuss its dynamical implications.17
Note that the flaring stellar disk is in contrast to a flat distribution or
a radially constant scale height that is commonly assumed in the literature.
The latter was proposed by [22] from analysis of their data. To explain
this claim of a constant scale height, [22] further proposed a specific rate of
radial fall-off in dispersion; namely, Rv /RD = 2, where Rv and RD are the
scale lengths for the exponential fall-off in the vertical velocity dispersion
and surface density with radius, respectively (as defined in Appendix A).
Both these claims by [22]: namely, that the scale height is constant, and

17
This section can be read as a largely stand-alone mini-review on this topic.

110
Rv /RD = 2; have now shown to be invalid in real galaxies, as shown by
observations as well as theory, as will be discussed in this section.

5.1. Physical basis for the radial variation in vertical thickness

In order to see the significance of a flaring disk, let us first consider some
general theoretical ideas on what determines the vertical scale height of the
disk at a given radius.
First, consider a single-component disk. A self-gravitating, isothermal
disk in pressure equilibrium obeys a vertical density distribution of form
sech2 (z/z0 ) ([12]; also, Section 3), where z0 = [⟨(vz )2 ⟩/2πGΣ]. Here, the
parameter z0 is a measure of the disk scale height of the sech2 distribution.
The value of z0 is set by the balance between the self-gravitational force and
the vertical pressure gradient (Section 3.1). For a general density distribution
(as for example, resulting for a multi-component disk plus halo case), the
HWHM of the stellar vertical distribution is taken to define the disk thickness
([16]; also, Section 4.2.2, and Appendix A). For a sech2 distribution, the disk
thickness (HWHM) is given by = z0 × sech−1 (0.5) = z0 × 0.88 (Appendix A).
Thus, for a one-component case, the disk thickness (HWHM) is proportional
to ⟨(vz )2 ⟩/Σ. This also indicates the formal functional dependence of the
disk thickness (HWHM) for a general density distribution, ρ(z).

5.1.1. Dependence on radial variation of stellar velocity dispersion

It is clear that for a one-component stars-alone disk, if Rv = 2RD , then
the scale height or the disk thickness for stars (which is ∝ ⟨(vz )2 ⟩/Σ, as
shown above, here both these parameters are for stars) remains constant
with radius. Physically, here the fall-off in pressure precisely compensates
for the fall-off in the disk gravity, hence the disk scale height remains constant
with radius. In the literature, the relation Rv /RD = 2 is routinely assumed,
following the claim by [22]; although there is no physical justification for this
relation to be valid. See, for example, the discussion in [29] about this point,
and also the discussion in Section 5.4. Indeed, for the Milky Way stellar disk,
the observed value of the ratio Rv /RD = 2.8(> 2) (see Section 5.3.1).
Note that if the velocity dispersion falls off more slowly than the above
required rate, namely if Rv /RD > 2, the pressure support falls slower with
radius than the gravitational force, hence the disk flares. That is, the pressure
support is higher at a given radius - than required for a constant scale height,
and this results in a flaring disk. This can be checked as follows:

111
  
exp(−2R/nRD ) R 2
HWHM ∝ = exp (1 − ) (5.1)
exp(−R/RD ) RD n
where Rv = nRD . For n > 2, the HWHM increases exponentially with radius.
Thus, for a single-component self-gravitating disk, any value of Rv /RD > 2 re-
sults in a flaring disk. Physically, for a one-component, self-gravitating disk,
for Rv /RD > 2, the pressure support falls slower with radius than the grav-
itational force, hence the disk flares. The higher the value of Rv /RD (> 2),
the higher is the flaring.
In a realistic multi-component disk plus halo system, however, no such
simple relation exists between Rv and RD that would result in a constant
thickness, or otherwise a value that would give a flaring disk. In the coupled
case, the gas and the halo exert a considerable gravitational force on the
stellar disk. In this case, the rate of variation of stellar disk thickness would
depend crucially on the value of Rv /RD (> 2) [35, 91]; and also on the surface
density of gas and the amount of halo mass, and their radial variation.
Thus, it is not straightforward to predict the limiting value of Rv /RD
above which flaring would be seen in a multi-component disk plus halo model.
However, it is easy to see that because of the additional gravitational force
due to gas and the halo, a higher pressure gradient in the stars is required
at a given radius, R, to keep the disk in pressure equilibrium with the same
resulting vertical scale height. This is because in the coupled case, the addi-
tional gravitational force due to gas and halo results in a smaller vertical stel-
lar scale height for the disk in pressure equilibrium; than in one-component
stars-alone case, at each radius, as shown for example for the Milky Way (see
Fig. 6, panel for stars). Further, the additional gravitational force due the
gas and halo is more important at larger radii. Hence, the pressure has to fall
more slowly with radius than in the one-component case to get a constant
scale height. Hence, in general in a multi-component disk plus halo system,
one needs higher dispersion at large radii, corresponding to Rv /RD > 2 to
even get a flat distribution. This was shown in the context of modeling of
two galaxies: NGC 891 and NGC 4565 by [35] (see Fig. 16) for NGC 891).
For the same reason, to get flaring in a multi-component disk plus halo case,
an even higher value of Rv /RD (> 2) is required than in the one-component
case.
In other words, for the same value of Rv /nRD > 1 where n > 2, the flaring
seen in the stars in the multi-component disk plus halo system is less than
in the stars-alone case (as shown for the Milky Way, see Fig. 6). Because

112
of the constraining effect of gas and halo, the stellar scale height does not
flare exponentially at large radii as it does in a single-component, stars-alone
case; instead, it shows a controlled linear increase with radius, for the same
observed Rv /RD = 8.7/3.2 = 2.7 > 2, for the Milky Way (see Fig. 6, panel
for stars). The stellar scale height values are lower for the multi-component
disk plus halo case than for the one-component stars-alone case. These lower
values agree with the trend from observational data from COBE [16].
A related point shown from the modeling of NGC 891 and NGC 4565
[35] is that a higher value of Rv /RD (> 2) results in a higher flaring (see Fig.
16). This is because, other parameters being the same for a given galaxy, a
higher value of Rv /RD (> 2) means a slower fall-off in pressure. This results
in a higher pressure support at a given radius and hence results in higher
flaring (see Fig. 17). This is also true for for a one-component case, as can
be seen from Eq.(5.1).

5.2. Origin of sharp flaring in the outer disk

In addition, there is another factor that can cause a sharp flaring of stellar
disk at large radii; namely, when the stellar dispersion falls sufficiently and
saturates at the gas dispersion values, as argued on physical grounds ([19];
also see [122]), as discussed next.
In the Milky Way, the stellar vertical velocity dispersion falls exponen-
tially with radius, with a scale length, Rv = 8.7 kpc ([56], see Section 4.2.3).
The velocity dispersion in external galaxies is difficult to observe and had
been measured observationally for only a few galaxies [57]. This situation is
now changing for the better over the last ∼ ten years due to the wide avail-
ability of IFU 2-D data on stellar kinematics in galaxies (see Section 2.1.4,
and Section 8).
In their study of the Milky Way focusing on the outer disk, [19] proposed
that beyond a certain radius when the stellar velocity dispersion falls close to
the value of the HI velocity dispersion, the stellar dispersion does not fall any
further; rather, it saturates to the gas dispersion value (see Section 4.2.3).
This is based on the physical argument that the velocity dispersion of stars
cannot be lower than the dispersion within the gas from which they form.
Beyond this radius where the stellar dispersion saturates, the stellar surface
density still continues to fall exponentially. Hence, the resulting vertical
stellar scale height would show a sharp flaring.
This physically motivated saturation in the stellar velocity dispersion was
proposed and employed in the theoretical model by [19]. This work showed

113
that for the observed input parameters for the Milky Way, the saturation in
the stellar velocity would occur at R ∼ 17 kpc. Further, their model results
indeed show that the resulting stellar scale height flares sharply beyond this
radius, as argued above, see Fig. 12.
Interestingly, using recent observations from Gaia and APOGEE, [122]
confirmed that the vertical stellar velocity dispersion saturates to ∼ 10 km
s−1 in the outer Galaxy (see Fig. 1 in [122]). In fact they predicted that
consequently the stellar disk would flare beyond this point. The physical
argument used by [122] to explain the observed saturation of the stellar
velocity dispersion was identical to the one proposed by [19], namely that the
stellar velocity cannot fall below the gas dispersion value, and hence saturates
to the non-zero velocity dispersion of stars at birth, which corresponds to the
dispersion within the gas clouds. Note that, the flaring of stellar disk beyond
the point of flattening in stellar dispersion, as predicted by [122], was exactly
already shown by the theoretical model results of [19].
Applying the same physical argument of saturation of stellar velocity
dispersion at large radii in the theoretical study of a low surface brightness
(LSB) galaxy, UGC 7321, [91] found that in this galaxy the stellar velocity
dispersion saturates at ∼ 7 kpc. Their results showed that, indeed, the model
stellar scale height shows sharp flaring beyond this radius (see Fig. 5 from
[91]; also, see Section 5.3.2 for details).
We predict that such onset of sharp flaring of stellar disk beyond the tran-
sition radius is expected to be seen in all galaxies. Recall that typically the
gas velocity dispersion tapers off and is constant at ∼ 7 km s−1 at large radii
in most galaxies (Section 4.2.3). The transition radius is set by the central
stellar velocity dispersion, the rate of fall-off in stellar velocity dispersion,
Rv , and the gas velocity dispersion, in the outer parts in a galactic disk (as
discussed by [91]).
The HI velocity dispersion as a function of radius has been measured for
a large number of galaxies in the THINGS (The HI Nearby Galaxy Survey)
[174]. The stellar velocity dispersion values including the central dispersion
can be extracted from the various IFU surveys although it is a challenging
task (see Section 8 for details). In principle, therefore, it would be straight-
forward to apply this result to more galaxies, and thus check if sharp flaring
of a stellar disk in the outer regions of disks, as predicted here, is indeed a
common result.

114
5.3. Application to galaxies: Theoretical modeling and observational results
Here we apply the above ideas about Rv /RD , and the radial variation in
stellar scale height, to real galaxies, to better understand the results from
observations and theoretical models given in the previous sections. We also
give additional evidence for flaring of stellar disks from observations and
simulations.

5.3.1. The Milky Way

Consider the Milky Way first, for which the input parameters for the
multi-component disk plus halo model are well-known observationally. In
fact, of all the galaxies studied so far; the input parameters, including the
variation in the stellar velocity dispersion with radius, are best known for the
Milky Way. In the Milky Way, the vertical stellar velocity dispersion is known
to fall exponentially with radius, with a scale length, Rv = 8.7 kpc (Section
4.2.3). Using this and the other input parameters, a single-component stellar
disk model in pressure equilibrium shows an exponentially flaring behaviour
([16]; also Fig. 6, panel for stars, dashed line). Inclusion of gas and halo
gravity results in a more moderate, linear increase up to R=12 kpc (Fig.
6, panel for stars, solid line). Yet even in the latter, more realistic case,
the scale height clearly shows flaring. Thus, flaring of a stellar disk is a
robust result. The resulting model rate of a moderate rise in scale height
with radius, between the radial range R= 5-10 kpc, is 24 pc kpc−1 . This was
shown to be in good agreement with the observations known at that time;
namely, the linear gradient deduced from COBE and SPACELAB 2 data
(see Section 4.2.4). The flaring behaviour is now confirmed by more modern
observations, as discussed later in this subsection.
In retrospect, given that the observed value of Rv /RD = 8.7/3.2 = 2.7 is
> 2, a flaring disk would be expected in the Milky Way; as seen from the
discussion given at the beginning of this section. However, to our knowledge,
this cause and effect correlation has not been generally noted in the literature.
It is not often realized that the claims by [22]; namely, that the scale height
is constant, and Rv /RD = 2, are not valid for a stars-alone disk even in the
Milky Way, for which the observed values of these parameters are known. In
fact, because the observed value of Rv /RD for the Milky Way is 2.7 > 2, the
stars-alone disk would flare exponentially. It is the presence of additional
gravitational force due to the gas and the halo in the coupled case that
decreases the stellar flaring to a moderate value (for R < 12 kpc), as seen
above.

115
 In the outer Galaxy, the physically motivated assumption of saturation
of stellar velocity dispersion (as discussed above), occurs at ∼ 17 kpc. The
theoretical model results indeed show a sharply flaring stellar disk beyond
this radius, as discussed above ([19]; also, see Fig. 12 here). The scale height
increases by a factor of 2.2 between R= 16 kpc and 22 kpc (see Table 2,
column 4 from [19]). The overall phenomenon of flaring in the outer Galaxy
as predicted by [19] is confirmed by observations of number counts of stars
(see Section 4.2.6 for details). The flaring of the stellar disk in the Milky
Way has recently been further confirmed by studies of number counts of
different tracers, such as: red clump stars [175, 176]; young O-B stars [177];
and supergiants [178]. In fact, all these papers, including the ones mentioned
in Section 4.2.6, simultaneously detect a flare and a warp in the Milky Way
stellar disk (also, see [39]).
The value of flaring measured in these papers (for R < 16 kpc) is some-
what higher than the model results. This discrepancy could arise because of
the incorrect deduction of warp, or contamination by thick disk stars – as was
argued by [127] for a similar excess in the measurement of stellar flaring seen
by [41]. Interestingly, using the vertical metallicity gradient of stars in the
Milky Way with the RAVE and Gaia data, and comparing with simulations;
[179] conclude that these stars have formed in a flaring stellar disk.
Thus, a flaring disk appears to be a generic result for a typical galactic
disk in hydrostatic equilibrium, and when the observed parameters are used.
It should be stressed that the flaring is not due to the multi-component
treatment; in fact, the reverse is the case. Keeping all the other parameters
constant (including Rv /RD to be equal to the observed value); the additional
gravitational force due to gas and halo in the coupled case actually decreases
flaring. That is, it leads to a more moderate rise with radius in the stellar
scale height. This was shown earlier for the Milky Way (see Section 4.2.4,
and the discussion above, as well as the next paragraph).
The flaring of the stellar disk is also seen in theoretical models which take
account of more detailed physics: such as, the treatment with complete Jeans
equations as applied to stars-alone case (Model B in [20]); and the treatment
that takes account of the observed non-isothermal stellar velocity dispersion
for the multi-component disk plus halo case [21] – both applied to the Milky
Way. Interestingly, a stars-alone, thin, isothermal disk in pressure equilib-
rium for the observed parameters shows a flaring of scale height (HWHM)
by a factor of 13.6 between R= 4 to 22 kpc (see Table 2 from [19]); while
the complete, Jeans treatment (Model B from [20]) shows a smaller increase

116
in scale height of a factor of 6.2. In comparison to these two cases; in the
multi-component disk plus halo case, the flaring is further reduced to a factor
of 3.4, due to the gravitational force of gas and halo. In this case, the stellar
disk flares by a smaller factor of 3.4 between R= 4 to 22 kpc (Table 2 from
[19]).
The recent observations for the Milky Way provide a different picture
of the radial variation of stellar kinematics, which nevertheless strengthens
the feature of a flaring disk. [65] analyze the detailed information about the
motion of a large sample (∼ 65000) of stars, located within a galactocentric
radial range of 4 to 13 kpc, as observed from APOGEE and Gaia. They show
that z velocity dispersion for stars shows a clear increase with radius beyond
9 kpc, for all ages of stars (see Fig. 8 from [65]). This is surprising, and
opposite to the exponential fall with radius, as observed in earlier studies
for the Milky Way [56]; and, also in external galaxies [57]. The physical
reason for the increase of stellar velocity dispersion with radius is not well-
understood. It could be due to a tidal encounter with a satellite galaxy. In
any case, it is easy to see that in this case, the scale height would increase
with radius. This prediction is easy to check for a one-component stellar
disk, where the thickness z0 would be ∝ (σz )2 /Σ (both quantities for stars) –
which would steadily increase with radius. At each radius, the resulting scale
height would be higher than the values obtained earlier (using the previously
observed radial fall-off in velocity dispersion with Rv =8.7 kpc =2.7 RD ). The
new kinematic data indeed shows that the disk flares considerably [180, 181],
which supports the above prediction.

5.3.2. UGC 7321

The scale heights of the stellar disk have been measured observationally
in the edge-on LSB galaxy, UGC 7321 up to 6 kpc. This shows a moderate
flaring of a factor ∼ 2.5 upto 6 kpc (∼ 3RD ). Using this information on
scale heights as a constraint to the multi-component disk plus halo model,
the value of the rate of fall-off, Rv /RD , was obtained to be =3.2 [91]. The
same value of Rv /RD was applied to theoretically predict the scale height
beyond this radius, and the model scale height was shown to be rising. Again,
the physically motivated idea of saturation of the stellar velocity to the HI
dispersion (as proposed by [19]) was applied here. The saturation occurs at
∼ 7 kpc in this case. As expected (see the discussion in Section 5.2), the
resulting model stellar scale height shows sharp flaring beyond this point
(see Fig. 5 in [91]; also, see the discussion in Section 4.3.2). Thus, both the

117
Milky Way and UGC 7321 show a similar pattern of flaring of the stellar disk,
with the onset in sharp flaring occurring where the stellar velocity dispersion
saturates to the value of the gas dispersion.

5.3.3. NGC 891, NGC 4565

Two other galaxies, NGC 891, and NGC 4565, were also shown to have
a flaring stellar disk. This was shown from the correct interpretation of the
data of these galaxies ([35]; also, see Section 2.1.1 and Section 4.2.7). On
a careful re-examination of the analysis by [22], [35] found that the data
by [22] actually shows a significant scatter. A correct interpretation of this
scatter indicates a moderate increase in the stellar scale height with radius,
corresponding to a factor of ∼ 2 − 3 within the optical disk. [35] next applied
the multi-component disk plus halo model to these galaxies, and used this
deduced variation in scale height to constrain the rate of radial fall-off of
stellar velocity dispersion. The ratio Rv /RD was thus shown to lie between
2-2.5 for NGC 891 and between 2.5-3 for NGC 4565. An independent study,
by [36], also showed that for these two galaxies, a flaring disk gave a better
fit to the intensity profile data of [22] (for details, see Section 2.1.1).

5.3.4. Other galaxies

It was already pointed out by [29], based on a sample of 48 galaxies, that
the data shows a small variation in scale height with radius; but this result
was largely ignored. More recently, there has been further observational
evidence for flaring of stellar disk in a number of galaxies: see [182, 183];
including an ultra-thin Galaxy UGC 11859 [184].
Recently, [185] have analyzed the observed data for 46 spirals from the
Spitzer Survey of Stellar Structure in Galaxies (S4 G) catalogue. By fitting
the data by an exponential and sech profile for the radial and vertical mass
distribution respectively, and following the procedure as in [22], it has been
been shown that the stellar disk flares by a factor of few within the optical
disk in all the galaxies considered [185].
However, there are several caveats regarding this analysis and the result.
First, the above method uses the integrated intensity along the line of sight,
given by (Eq. (2.2)) as proposed by [22]. This derivation implicitly assumes
the scale height to be constant with radius, which is not warranted, as pointed
out in Section 2.1.1. The work by [185] employs two internally inconsistent
assumptions: of a constant scale height, as well as a scale height that varies

118
with R. Thus the scale height values, obtained by [185]), by fitting the
image along cuts at different radii will carry over this error (see the general
discussion in Sec 2.1.1; also, see [182]). Thus the values of scale heights
as obtained by [185]; are not accurate; however, the trend of a flaring disk
obtained in all galaxies by them will probably remain valid.
Another problem is that the approach by [185] involves a simple fitting
of sech or sech2 functions unlike the more accurate approach by [160]. [160]
had used the multi-component disk plus halo model to obtain physically
motivated disk vertical density profiles for thin and Thick disks, and thus
extract more accurate disk scale heights from the same S4 G data (Section
4.5.1).
Finally, another issue with their approach is that the results given by
[185] are for a single disk fitted to the data; so the scale height obtained
represents a cumulative fit to the thin and Thick disks in each galaxy.

5.3.5. Evidence of flaring from numerical simulations

There is a growing evidence for a flaring stellar disk from results from sim-
ulation studies of galaxies, especially when the artificial condition of constant
thickness used earlier is relaxed. Note that simulations data is amenable to
easily being fit by a given functional form of the density distribution to see
how well the fit is; and to obtain the corresponding HWHM values. Earlier,
the flaring effect may have been missed in the simulations results; perhaps
because it was not looked for, or the resolution in the earlier simulations was
not sufficient.
One early evidence for a flaring stellar disk comes from simulations of
galactic dynamics of the Milky Way by [186], which showed a small increase
in stellar scale height with radius. Recently, in an important, systematic
study of the TNG50 simulations data, [129] have fitted the TNG galaxies by
a function of the form sech(2/n) (z/z0 ) (proposed by [14]) to determine the
scale height. They show that the flaring effect is robust and is seen whether
the data is fitted by vertical density profile that has a sech or a sech2 form.
It is also seen when a half-mass scale height (also defined by [19]) was used.
However, a word of caution about the results by [129] is in order. Recall that
the self-consistent density distribution in a realistic multi-component disk
plus halo model as applied to the Milky Way [19], and to UGC 7321 [91],
shows that the parameter n obtained is not unique. Rather, it is a function
of the range of z, ∆z, over which the fit is made. Thus, n obtained this way

119
is not a well-defined, unique number (see Section 4.2.6; and Fig. 13). Since
[129] fit a single function to the entire z range; therefore, the resulting n as
well as z0 they obtain are not well-defined. However, the trend of a flaring
stellar disk obtained by them probably remains valid.

5.4. Discussion
To conclude from the above applications, a flaring stellar disk appears to
be a generic feature in galaxies in both the High surface brightness (HSB)
galaxies, such as the Milky Way; and the Low surface brightness (LSB)
galaxies. As shown above, the stellar disk flaring is directly tied up with the
value of Rv /RD , which has to be > 2 for the disk to show an increasing scale
height with radius. However, what decides the value of Rv is not understood;
and in fact, this question has not been discussed much in the literature and
needs to be explored. The value of Rv would be decided by how the heating of
stellar velocity dispersion varies with radius: the various sources of heating
being the gravitational encounters with gas clouds, spiral arms and a bar
within a galaxy; and tidal encounters with passing satellite galaxies (see
Section 1). For recent studies of this topic, see [23, 24]. Given the diverse
ways in which the value of Rv could be affected, it is clear that there is
no physical justification for using Rv /RD = 2 as is routinely done in the
literature. Also, as discussed above, a few galaxies for which RD and Rv
have been measured observationally, give a value of Rv /RD > 2.
Flaring of HI gas disk
For the sake of comparison, consider the scale height variation with radius
for the HI disk. It is easy to see that the gas component would show flaring.
For the gas responding to the stellar potential alone (Eq. 3.9), the gas scale
height, (z0 )g ∝ (σg2 /Σs ), where σg is the gas velocity dispersion and Σs is the
stellar disk surface density (see Table 1). Since the stellar surface density
would fall with radius for any realistic disk, and the gas velocity dispersion
is nearly constant with radius (see the discussion in Section 4.2.3, we expect
the gas to flare with radius (see Section 3.2). The HI gas in the Milky Way
indeed shows flaring at large radii ([42, 111, 110]; also, see Section 4.2.6, and
Section 6). At large radii, the gravitational force of the halo dominates, which
decreases the gas flaring to a moderate value ([19]; also, see Section 4.2.6).
The HI gas disk flaring at large radii is a common feature of spiral galaxies.
This feature can be used to constrain the dark matter halo parameters, by
theoretical modeling using a multi-component galactic disk plus halo system
(see Section 6 for details).

120
 On the other hand, in the inner regions of the Galaxy, at R < 8.5 kpc,
the gas scale height is observed to be nearly constant [109, 67]: this was a
long-standing puzzle. To solve this problem, gas gravity was brought into the
picture; and this was one main motivation to propose the multi-component
disk plus halo model ([16]; also, see the discussion in Section 3.2, and Section
4.1).
Disk Flaring: Why it should be included in future studies
Thus, there is clear and growing evidence, both observationally and the-
oretically, that the stellar scale height is not strictly constant; rather, it in-
creases moderately, by a factor of few, within the optical disk. Despite this,
the assumption of a constant scale height is commonly used in the literature,
perhaps because it simplifies the treatment. For example, the assumptions
of a constant scale height and Rv /RD = 2 are often used in model building
in the literature, including in numerical simulations, as the basic state of
a galaxy (e.g., [187, 188]); as well as in interpreting observational data, as
in the analysis of 2-D kinematic data [58]. We stress that adopting these
wrong assumptions is not justified physically and could lead to erroneous re-
sults. For example, the early and much-cited work on 2-D kinematics using
IFU data by [58] has a constant disk scale height as a built-in assumption.
However, since the actual scale height increases radially, the surface density
values they deduce from their measurement of vertical dispersion would not
be accurate. The true surface density would be lower than the values they
deduce.
We urge that future theoretical studies of vertical structure of galaxies
including N-body simulations, and also analysis of observations, should ex-
plicitly incorporate a flaring stellar disk for a correct understanding of the
dynamics and evolution of a galactic disk.

5.5. Dynamical implications, and future directions

The finite height of a galactic disk can affect many dynamical features,
including the in-plane dynamics of a disk: such as, the criterion for stability
against axisymmetric perturbations of a disk of finite height [189, 163, 190],
and star formation [165]. The vertical thickness of a disk also tends to
dampen or quench the swing amplification of non-axisymmetric perturba-
tions in a disk [191]. That is, a finite disk thickness leads to suppression of
spiral arms. A finite disk thickness also leads to decrease the lifetime of a
spiral density wave because it increases the group transport velocity [192]. A

121
finite height of the disk thus makes it more stable against instabilities. The
actual value of height will affect the effectiveness of these phenomena. Due
to flaring of the stellar disk, these effects of a finite disk thickness would be
more important at large radii. For example, a disk would resist the onset
of non-axisymmetric instabilities at large R, and the spiral arms would be
shorter-lived in the outer disk. Thus a flaring disk makes it harder for a disk
to support instabilities within it at large radii.
On the other hand, a flaring disk itself is more susceptible to being tidally
distorted at large R. The multi-component system is harder to be disturbed
than the stars-alone disk [90]; this needs to be checked by numerical simu-
lations. As discussed earlier, the constraining effect of gas and dark matter
halo reduces the flaring of the stellar disk. Thus, overall, a multi-component
disk and especially a dark matter halo, tend to stabilize a stellar disk against
distortion by controlling the flaring to be of a moderate level.
Future theoretical studies, including N-body simulations, should take ac-
count of the flaring stellar disk, in order to portray the correct dynamical
picture.
Future directions for study of a flaring stellar disk
Firstly, and most importantly; new and accurate observational data, and
better analysis of the data, are needed to quantitatively better define the
phenomenon of a flaring disk. To study the stellar scale height as a function
of radius; a proper, iterative modeling of the observed intensity profiles, I(z),
that takes account of the variation in thickness with radius is needed (see
Section 2.1.1). This will directly give the measured stellar scale height as a
function of R.
Ironically, even when a radial variation in scale height is claimed to be
measured, as in the recent work by [185] involving analysis of S4 G data on
galaxies, the equation used by them for fitting the data (Eq. (2.2), which was
derived by [22]) implicitly assumes a radially constant scale height. Hence
the values of scale heights as measured by [185] are not strictly correct, as
discussed above in Section 5.3.4.
Alternatively, the stellar scale height values can be calculated theoreti-
cally, if the necessary input parameters are available from observations. The
wide availability of IFU kinematic data for 2-D stellar kinematics in external
galaxies (Section 2.1.4) can now allow for a direct measurement of the stellar
velocity dispersion as a function of R. Note, however, that it is a painstaking
task to extract this information from the data (e.g., [60]; see Section 8). If
other physical parameters such as disk surface density are known from obser-

122
vations, then the multi-component disk plus halo model could be applied to
theoretically predict the radial variation of stellar vertical scale height. This
can check if flaring is indeed a general feature of a galactic stellar disk.
In addition, the various dynamical implications of a flaring disk, as dis-
cussed above, need to be studied in future.

6. Density profile and shape of dark matter halo

In this section, we consider an important application of disk vertical struc-
ture as a diagnostic to determine the parameters of the dark matter halo in a
spiral galaxy.18 We highlight how the observed gas vertical scale height can
be used as a complementary constraint, along with the observed rotation
curve, to quantitatively determine the parameters such as the density profile
and shape of the dark matter halo. This approach has been applied to a
number of galaxies, including the Milky Way, as discussed here.
The shape of the dark matter halo is an important parameter to under-
stand its formation and evolution (e.g. [193] for an early discussion on this);
and also because it can strongly affect the dynamics of the disk embedded
in it, including features such as warps and bars in the disk. However, the
properties of the halo shape are not yet well-understood. The deduction of
both, the density profile and the shape of the halo, using observations of
tracers is an important problem; but so far it has not received the focused
attention that it deserves.

6.1. Background on dark matter halos in galaxies

First, we give a brief background on the dark matter halos in spiral galax-
ies, to put this topic into a proper context. It is now well-accepted that a
typical spiral galaxy is located in an extended, massive dark matter halo
(e.g., [2, 194]). Within the extent of the optical or stellar disk of a galactic
disk, most of the mass is in baryons. The fraction of dark matter increases
with radius; such that at large radii, beyond the optical disk, most of the
galaxy mass is in dark matter and not baryons (see [195]; and, also the dis-
cussion in the next paragraph). The dark matter halo is believed to play an
important role in the formation, evolution and dynamics of a galaxy [194].
Despite being the dominant mass component of a galaxy, even the basic
physical parameters such as the mass, size, radial density profile, and the

18
This section can be read as a largely stand-alone mini-review on this topic.

123
shape of the dark matter halo are not yet well-understood. The dark matter
halo thus remains one of the most important, unsolved problems in the study
of galaxies. Since the dark matter halo interacts gravitationally with the
disk, the halo parameters could in principle be determined by its effect on
the disk, which can be observed. In fact, this was precisely how the existence
of dark matter halos in galaxies was deduced from the observed, extended,
nearly-flat rotation curves that did not show a Keplerian fall-off [196, 84,
197, 85, 198, 199]. This implied the existence of more gravitating mass than
can be accounted for by the visible matter. These studies of the rotation
curves of galaxies showed the existence of a dark matter halo whose fractional
mass content increases with radius, and which dominates in the outer parts
[199, 200, 201, 195]. For a summary of dark matter halos in galaxies as
deduced from the observed rotation curves in galaxies, see [4, 202].
For a disk in rotational equilibrium, the rotational velocity is set by the
gravitational force due to the disk, plus halo, (and the bulge) balancing the
centrifugal force. In these studies, the halo is assumed to be spherical or a
slightly flattened oblate spheroid, for simplicity. Thus, the rotational velocity
is a good tracer of mass distribution, and hence the potential in the plane of
the disk. Interestingly, the rotation curve is weakly dependent on the halo
shape [203], so it is not a sensitive indicator of the mass distribution normal
to the galactic plane. One needs to consider a tracer distribution extended
normal to the galactic plane to study the mass distribution normal to the
galactic plane. As discussed next, the HI gas scale height is an excellent
tracer of the gravitational potential gradient, and hence the shape of the
halo, normal to the plane, in the region close to the mid-plane.

6.2. Motivation for using HI gas scale height as a constraint

In a typical spiral galaxy, the HI gas extends several times (∼ 1.5 − 2
times) farther out than the optical or stellar disk (e.g., [204]). The optical
or stellar disk size as observed is typically ∼ 4 − 5 RD [22, 8, 205]. It is well-
known than the HI gas scale height flares in the outer Galaxy ([42, 111, 78];
also, see Sections 3.2,4.3.2,5). For HI gas in hydrostatic equilibrium, a possi-
ble cause of HI flaring is that the net gravitational force acting perpendicular
to the disk decreases with radius, while the velocity dispersion of HI gas is
observed to remain nearly constant with radius (Section 4.2.3). The contri-
bution to the net vertical gravitational force comes from stars, gas and the
dark matter halo; with the halo dominating in the outer disk. Thus the halo
plays a major role in determining the vertical disk structure in the outer

124
disk, especially beyond the optical disk. The HI gas vertical scale height is
a sensitive indicator of the the vertical gradient of the potential close to the
mid-plane, and hence is the best tracer to study the vertical mass distribu-
tion, and the shape of the halo, close to the mid-plane. We focus on this new
approach in this section.
The model HI scale height depends critically on the choice of the HI
velocity dispersion (Section 4.2). Typically this is observed to be a constant
at ∼ 8 km s−1 in the inner parts of a galactic disk, then tapering off to 7
km s−1 in the outer parts – see the discussion in Section 4.2.3); also, see
[174] for a measurement for the THINGS (The H I Nearby Galaxy Survey)
sample of galaxies. The actual value of gas velocity dispersion used in the
model calculations is taken from the observed value for each galaxy under
consideration, as discussed later in this Section.
The rotation curve is only weakly dependent on the halo shape [203].
Therefore, the observed HI gas vertical scale height, especially in the outer
disk, can be used as an additional and complementary constraint – in addition
to the observed rotation curve that mainly traces the planar mass distribution
– to determine the density profile and the shape of the halo. Using the above
two independent criteria on an equal footing allows one to better constrain
the 3-D mass distribution of the dark matter halo; and hence the shape of
the halo. This is a novel approach and has been used to determine the halo
parameters (e.g., [206, 207, 78]). This is an important application of the disk
vertical structure, as we discuss in this Section. In contrast, normally only
the rotation curve is used to determine the radial mass distribution of the
dark matter halo halo in a galaxy (e.g., [199, 208]).
Measuring the HI vertical scale height is difficult ([144]; also, Section 6.5),
so the few galaxies for which the HI scale height values are known, have been
studied so far to obtain the halo parameters using this approach.
We note that other tracers have also been used in the literature to study
the halo parameters, we will discuss these briefly in Section 6.7.
Radial density profile and shape of dark matter halo: values tried
The halo parameters, namely its radial density profile and shape, can
be obtained from numerical simulations, or from analysis of observations of
visible tracers such as stars or gas. Generally the rotation curve alone is
used to trace the halo parameters. For a particular choice of radial density
profile of the halo adopted, the de-composition of the observed rotation curve
in terms of the contributions from the disk, bulge and halo allows one to

125
constrain the halo parameters for the halo profile adopted.
Numerical cosmological simulations give a centrally peaked, NFW (Navarro-
Frenk-White) radial density profile for a dark matter halo [139]. This is often
used as a choice of halo density profile to model the observed rotation curve
in terms of the contributions by the stellar disk, gas disk, and the dark
matter halo, to constrain the halo parameters [208, 209, 210]. Once high
resolution rotation curves became available and it was possible to model the
rotation curves close to the centre, it became clear that the dark matter halo
is actually better fitted with a pseudo-isothermal distribution with a flat,
constant-density core [208]. This is particularly true in the case of late-type,
or low surface brightness (LSB) galaxies. In the literature, both choices are
often used to model rotation curves [146, 209]. In fact, some studies [211]
have shown that both give equally good fits. Studies of dwarf galaxies often
use the density distribution given by [212] for the halo.
Interestingly, studies in the literature that use the HI gas scale heights and
the rotation curve together as tracers for the halo potential - which is the case
we will focus on - have mostly used the pseudo-isothermal density distribution
with a flat core, as it is found to give a better fit to the observations.
Shape of the dark matter halo
Cosmological N-body simulations typically give triaxial-shaped halos, with
a ̸= b ̸= c, where a and b are the semi-major and semi-minor axis in the
galactic plane and c is the semi-major axis along the vertical axis (or, the
vertical direction w.r.t. the plane). In other words, a,b,c are measured along
the Cartesian axes X, Y and Z respectively. Typically, the disk lies in the
plane defined by a and b axes. In simulations, the axis ratio of the halo,
or the shape, is found to vary with radius, and the shape is also seen to
evolve with time [213, 214, 215, 216]. Simulations with baryonic feedback
from the disk give a spherical or an oblate-shaped halo in the inner regions
([217, 218, 219]; also, see the discussion in [220]).
In most studies of real galaxies, the triaxial shape is not taken into ac-
count, with some exceptions (e.g., [221, 222]). For most theoretical studies
dealing with real galaxies, as well as for analysis of observational data of
tracers; typically, a spherical (a = b = c), or a spheroidal shape is adopted
for simplicity: where c < a denotes an oblate spheroid, and c > a denotes a
prolate spheroid. Here onwards, we will refer to the quantity, q = c/a, the
ratio of the vertical-to-planar axes, as the halo shape (normal to the galactic
plane); so that q < 1 and q > 1 corresponds to an oblate and a prolate

126
spheroid, respectively. This is the quantity we want to trace using the HI
gas scale heights.
The shape measured by analysis of observations appears to depend on the
tracer used, and may be sensitive to or reflect the physical region where the
tracer is located, as will be discussed in Section 6.7. Due to the limitation of
resolution of data, typically a constant shape across the entire halo is used
in the literature, for simplicity.
In the work discussed in Section 6.3, the dependence of rotation curve on
the shape (albeit small) is taken into account to match with the observed
rotation curve. For this work, the generalized four-parameter halo density
profile (see Eq.4.14) as given by [138] is used, see Section 4.3.2 for details.
The shape of the halo in the plane (= b/a) will not be discussed, though
we mention it here briefly for a complete picture. The planar shape of the
halo can be determined as follows: by Fourier analysis of the observed isopho-
tal shapes of galaxies in the near-IR [223, 224, 225]; by Fourier analysis of
HI surface density maps [226, 227]; by Fourier analysis of the HI velocity
fields of galaxies [228]; and, by measuring the asymmetry in the HI rotation
curves [229]. These studies show that, typically, dark matter halo in a spiral
galaxy shows azimuthal asymmetry of type m = 1 (lopsidedness) and m = 2
(ellipticity) of a few percent amplitude each, in the disk plane.

6.3. Dark matter halo traced using HI gas scale height: Self-consistent ap-
proach
The pioneering idea on obtaining the halo parameters including the halo
shape, q, using the observed rotation curve and HI gas vertical scale heights
as two independent and simultaneous constraints was proposed in a model by
[120]. The above model was applied to NGC 4244 [206] and NGC 891 [207]
which gave highly flattened oblate halos with the axis ratios in the range
of 0.2-0.4. However, the model by [120] is not rigorous, and does not use a
self-consistent approach to obtain the vertical distribution of stars and gas
in the halo potential, as will be discussed later in Section 6.4.
The above idea of using the observed gas scale height and the rotation
curve as two independent constraints to determine the halo parameters was
extended; and applied in a systematic and accurate way to the Milky Way by
[78]. They obtained the self-consistent vertical distribution and hence the HI
scale height by using the multi-component disk plus halo model, and explored
a large range of density profiles and shapes (spherical, oblate and prolate) in
a systematic way. For this, the halo was modeled as a general, four-parameter

127
model by [138] (see [78] for details; also, see Eq. (4.14)). A similar approach
has also been applied to obtain the dark matter halo parameters of M31 and
UGC 7321. These are the galaxies for which observed HI scale height values
were available as constraints, as will be discussed later in this section.

6.3.1. Application to the Milky Way

The above careful treatment by [78] has already been described in detail in
Section 4.3.2, as an example of application of the multi-component disk plus
halo model for a thick disk case. The details of dark matter halo parameters
scanned and the best-fit values are given in the same section. [78] found that
while a prolate-shaped halo gave an overall better fit to the observed HI scale
heights, no single prolate shape was adequate to explain the observed flaring
over the entire radial range considered (R=8 to 24 kpc) (see Fig. 18).
Progressively more prolate halo
Instead, a halo that is increasingly more prolate with radius was indicated
that could explain the observed steep flaring of HI disk in the outer disk
(Section 4.3.2). This idea was taken up and worked out in detail by [140] in
a subsequent paper. This work is based on the ingenious use of the fact that
the mass of a spheroidal shell bounded by two surfaces m and m + δm (where
the parameter m defines an isodensity contour), is independent of the shape
of the spheroidal shell (qR ) at R [2]. Thus, if an original spheroidal shell with
radius R is distorted to be a prolate spheroidal shell by construction; then
the condition of constancy of mass of the shell gives the following condition:

qR ρR (qR ) = ρR (q = 1) (6.1)
where ρR (qR ) is the density along a prolate isodensity contour through R
and ρR (q = 1) is the density along the corresponding spherical contour with
radius R. The quantity qR denotes the vertical to planar axis ratio, which is
> 1 for a prolate-shaped halo (Section 6.2). Thus, the density of the prolate
shell at R will be lower by a factor qR . Hence, the vertical force near the
mid-plane will be lower. Recall that the self-consistent ρ(z) is determined
locally at a given R (see Section 4.2.1, and Section 4.3.1). The resulting
thickness of the disk in pressure equilibrium is strongly dependent on the
vertical gravitational force near the mid-plane at a given R ([18]; also, see
Section 4.2.5). Therefore, the resulting thickness for a prolate shell will be
higher. A small change in the shape of a shell of a given mass has striking
effect on vertical scale height since the mid-plane density decreases linearly

128
Figure 28: Calculated vertical scale height for the atomic hydrogen gas (HI) (solid line)
and the observed values (Wouterloot et al. 1990) (filled circles) vs. the radius R in the
outer Galaxy. The theoretical curve is the best-fit case and corresponds to a dark matter
halo which is increasingly more prolate. In the radial range studied, the halo is found to
be most prolate with the vertical-to-planar axes ratio, qR = 2, at R=24 kpc. Source:
Taken from [140]

as 1/qR . This could in principle explain the observed steep flaring of HI

if qR were taken to increase with radius in the outer disk, where the halo
contribution to the gravitational force becomes important. This was shown
by [140] who proposed and applied this idea to work out the variation of halo
shape with radius.
To start with, [140] consider a spherical, pseudo-isothermal halo as from
the Galactic mass model by [82], with the density profile ρ(R) as given by
Eq.(4.6), at z = 0. The density at a given R multiplied by 1/qR – the
reduction in density due to distortion of a shell to a prolate shape, as given
by Eq.(6.1) – gives the net reduced density in a prolate shell at a radius R.
This reduced value is used as the halo density in the coupled, joint Poisson-
hydrostatic balance equation (Eq. (4.12)) for the multi-component disk plus
halo model, in the thick disk case, to obtain the HI scale height at a given
R. Recall that the vertical structure is determined at each R.

129
Figure 29: Resulting prolate-shaped isodensity contours (on the R-z plane) of the best-fit
dark matter halo; with values of density (as one moves outwards) being 8.3, 5.2, 3.5, 2.5,
1.9 and 1.5 (in units of 10−3 M⊙ pc−3 ). These contours denote a progressively more
prolate halo in the outer Galaxy. This corresponds to qR , the vertical-to-planar axis ratio,
increasing with R as shown in the in the inset. This brings out the increasingly prolate
shape of the halo. Source: Taken from [140]

Each shell acts independently, and qR is a local property of each shell.

The radial variation of qR is modeled as a simple second-order polynomial
in radius. See Fig. 28 for the best-fit to the observed HI scale height data
obtained. This procedure constrains the variation in halo shape and gives qR
as a function of R. The halo shape is found to vary from a spherical shape at
R=9 kpc (by assumption), to an increasingly prolate halo with q(R) = 2 at
the maximum radius of R = 24 kpc studied – up to which the HI scale height
data is available. This method employing the shell-by-shell reconstruction
of the halo makes it possible to study the radial variation in halo shape. In
contrast, most other work in the literature until then had mostly modeled
the dark matter halo as a spherical or oblate halo of constant shape (see [140]
for a detailed discussion and the related references).
Fig. 29 gives the isodensity contours (on the R-z plane) corresponding
to the above increasingly prolate best-fit halo in the outer Galaxy; and the

130
inset shows the resulting qR vs. R. As shown above, an increase in qR from
1 to 2 corresponds to a decrease in mid-plane density of a factor of 2 (Eq.
(6.1)). Interestingly, since the resulting HI scale height depends non-linearly
on the mid-plane density; this moderate radial variation in the halo shape is
sufficient to explain the observed steep flaring of HI scale heights by a factor
of 6 between R= 9 to 24 kpc.
Interestingly, the resulting prolate-shaped nature of halo agrees with the
general trend seen in cosmological simulations of that time, which show a
preference for a prolate-shaped halo [213, 214]; except that, the simulations
show a smaller vertical-to-planar axis ratio. However, in simulations, the
axis ratio values are measured over a larger spatial scale of ∼ 100 kpc [213];
hence a quantitative comparison between these two is not meaningful. For
the current status of halo shapes in simulations, see Section 6.2. These give
an oblate shape as the preferred shape; while the observations give a spherical
or a slightly prolate shape as the preferred shape [220].
In another study, [230] also tried to fit the HI flaring with a prolate halo
by treating a multi-component disk plus halo model. They used the same
halo profile (Eq.(4.14)). They found that a constant-shaped prolate halo,
even with an axis ratio, q, as high as 64 does not fit the steeply flaring HI
observed in the outer disk. They state that this confirms the trend found
by [78], namely that the results from a single-shape halo do not fit the scale
heights over the entire radial range. Instead, [230] show that the best-fit
to the HI scale height data up to R = 40 kpc is obtained by considering a
spheroidal dark matter halo, and a dark matter ring in the Milky Way disk
plane. There is a caveat, however, about the scale heights obtained by [230].
For a constant mass halo, the halo central density, ρ0h , and the core radius,
Rc , are not independent as [230] have assumed; rather, these two parameters
are related to each other through the halo shape, q, as shown by [78]. Hence
the results for scale heights obtained by [17] for a particular halo shape tried
are not accurate. However, their conclusion about mismatch with HI scale
heights obtained for a single halo shape would probably remain valid.
An important point to note is that the actual value of the scale height
and hence the resulting shape, qR , obtained depends critically on the choice
of the gas velocity dispersion; which is not well-known [78, 140]. Hence, the
value of qR as constrained by [140] may not be robust. However, given the
sharply flaring HI scale height values observed beyond 18 kpc in the outer
Galaxy, they expect that the qualitative trend of a halo that is increasingly
more prolate with radius, will still hold good.

131
 While the observed HI scale height constrains the shape of the halo to be
increasingly more prolate with radius, what gives rise to such a halo shape
is not understood. For that matter, in general the physical origin of the halo
shape in a real galaxy has not been understood or even discussed much (see
Section 6.7). Irrespective of the origin of the halo shape that is more prolate
with increasing radius, it is interesting to ask whether such a configuration
can even be stable. This was studied by [231] who calculated the potential
for a progressively prolate halo analytically, by applying the potential theory
for ellipsoids [172, 2]. They calculated the differential acceleration due to
consecutive shells of different eccentricities - corresponding to the best-fit
halo values obeying the qR vs. R as obtained by [140]. They showed that the
net force due to this is small; hence, such a system is stable over the Hubble
time.
A prolate halo, in particular one that is increasingly more prolate at larger
radii, could have interesting implications for galaxy dynamics and evolution,
which needs to be explored further (also, see Section 6.6).
Asymmetric HI scale height distribution and halo shape
A somewhat subtle point is that the observed HI scale height distribu-
tion shows an azimuthal asymmetry [111, 110], being higher by a factor of
two in the northern Galactic hemisphere than in the south. This has been
modeled by [232] in terms of a multi-component disk plus halo model where
the halo potential has a built-in azimuthal asymmetry in the plane of type
m = 1 (denoting lopsidedness) and m = 2 (denoting bars and spiral arms).
Lopsidedness is observed to be common in disk galaxies [223, 233, 225]; with
a magnitude comparable to the observed values for the m = 2 features. The
origin of lopsidedness is not yet well-understood. However, one generic mech-
anism that has been suggested is that the halo is distorted to have a lopsided
shape due to an interaction with a satellite galaxy [234, 233]; and the disk
then responds to it [235, 236, 228, 237]. The observed asymmetric HI scale
height distribution can be modeled in terms of the self-consistent vertical disk
distribution as a response to an external halo potential that is asymmetric,
as shown by [232]. This gives the best-fit halo potential to be lopsided with
a fractional amplitude of 0.18 (for m = 1); on which is superposed a m = 2
potential of amplitude 0.19 – which is out of phase with the lopsided per-
turbation potential. This choice explains the observed N-S (North-South)
asymmetry in the HI scale heights naturally. Such halo distortions can be
generated during galaxy encounters or mergers [234, 233, 238]. [232] argue
that the best-fit asymmetric halo they obtain is likely to be produced during

132
hierarchical merging of galaxies in the λCDM cosmology; and the halo being
collisionless, such asymmetry could be long-lasting.
We point out that in the work by [232], the asymmetry in the HI scale
heights in the two halves of the Galaxy is modeled in terms of the response
of HI to the asymmetry of the potential (of type m = 1, 2) in the disk plane.
This is a different tracer compared to the planar isophotal shapes which are
also used to deduce the planar asymmetry of the halo potential (Section 6.2).
Alternatively, the azimuthal asymmetry in the HI scale heights may be
explained in terms of intergalactic accretion flows, as proposed by [239].
Due to better data being available for it, the Milky Way is the best-
studied galaxy regarding its halo properties, and using a variety of tracers
(see Section 6.7).

6.3.2. Application to M31 (Andromeda)

The above technique of using observed HI scale heights and rotation curve
as two simultaneous constraints to study the halo parameters, was applied to
the Andromeda galaxy (known as M31 or NGC 224) by [240]. This problem
is treated using the multi-component disk plus halo model, in the thick disk
limit (see Section 4.3; also, see [78]). M31 being an Sb galaxy, the bulge
forms a substantial mass component in the central region, and can affect the
rotation curve even in the intermediate radial range. Hence the bulge needs
to be included in the correct formulation of the equations. This modifies
the joint Poisson-hydrostatic balance equation used (Eq.(4.12)) slightly, in
that the r.h.s. also includes the term corresponding to the density of the
bulge, ρb . The observed rotation curve is flat or gently falling, hence the
net R term in the combined Poisson equation is taken to be zero, as in [78].
The corresponding three coupled, joint Poisson-hydrostatic balance equations
(Eq(4.13), modified by the inclusion of ρb ); are solved together numerically,
using the same procedure as in [16]. The input parameters for the disk
and bulge are taken from observations [240]. For the halo density, the four-
parameter dark matter halo model is used ([138]; or, see Eq.(4.14)). The
halo parameter-space values are scanned systematically to get the best-fit to
the rotation curve and the HI scale height data in the outer disk. In analogy
with the Milky Way, the region outside of 3RD (∼ 16 kpc for M31) is treated
to be the outer disk and used for this study. The model results are matched
with the observed HI scale height values from [241] and the observed rotation
curve (see [240]) in this radial range, to obtain the best-fit values for the halo
parameters. A flattened isothermal halo (q = 0.4), with a central density,

133
ρ0h = 0.011M⊙ pc−3 ; and a core radius, Rc = 21 kpc is found to provide the
best-fit.
However, the paucity of data on the HI scale height values – with only
three data points for HI scale height in the outer disk – and the large error
bars on these, means that the above result for the best-fit halo parameters
is not robust. Additional scale height data, especially in the outer disk,
is needed for this technique to constrain the halo parameters better. For
comparison, in the Milky Way, the HI data was available up to 24 kpc [42],
or ∼ 8RD (that is, eight disk scale lengths); while in M31, the HI data does
not extend beyond 26 kpc (or, ∼ 5RD ). It is well-known, and surprising,
that the two large, nearby galaxies show such different HI gas distributions,
with M31 being overall gas-poor.
[240] show that the best-fit halo obtained is highly flattened, with a shape
q = 0.4; and that such halos lie at the most oblate end of halos seen in
cosmological simulations at that time [213, 214]. [240] state that the observed
flat HI scale height distribution in M31 indicates a high gravitational force;
which constrains the best-fit halo as being oblate. They point out, however,
that a moderate variation in HI gas velocity dispersion – that is well within
the observational limits – can result in the best-fit halo being less flattened.
For example, a smaller velocity dispersion of 7 km s−1 (instead of 8 km s−1
adopted in the paper), gives a lower pressure support and hence a smaller
model scale height. To explain the observed scale height, the halo then
needs to be less oblate; or, be more spherical (q = 0.8 − 0.9). Alternatively,
assuming a small radial fall-off with radius in the gas dispersion, from 8 km
s−1 to 7 km s−1 in the outer disk, as observed in the Galaxy; also results in
a lower pressure support and lower model HI gas scale height. Hence a less
oblate halo is implied, with q ∼ 0.5 − 0.6, to explain the observed HI scale
height distribution. However, [240] find that in this last case, the resulting q
values are not constrained well. This is because the small value of model HI
scale height is now mainly sought by a change in the gas velocity dispersion,
hence the dependence on the halo shape is weak.

6.3.3. Application to UGC 7321

Another application of the above technique was to UGC 7321, which
is a low surface brightness, superthin galaxy [242]. The aim was to study
the dark matter halo properties of the low surface brightness (LSB) galax-
ies. These are known to be dominated by dark matter halo starting from
the smallest radii within their stellar disks, as seen from fitting of rotation

134
curves alone [243]. The two simultaneous constraints of rotation curve and HI
vertical scale heights were used [242]; the observed values for both of which
are known from observations [150]. See Section 4.3.2 for a brief discussion
about low surface galaxies and the observed parameters for UGC 7321. This
galaxy has insignificant amount of molecular hydrogen gas. The rotation
curve is assumed to be flat for simplicity. The corresponding coupled, joint
hydrostatic-Poisson equations for a multi-component disk plus halo model in
the thick disk limit (Eq.(4.13), with ρH2 = 0), are solved together numeri-
cally, using the procedure as in [16]. This gives the self-consistent vertical
density distribution for stars and HI.
For solving the above coupled equations, the four-parameter halo po-
tential is used ([138]), or see Eq.(4.14)). The halo parameters are scanned
systematically and the resulting model rotation curve and the HI scale height
values are matched with the observed values, so as to get the best-fit values
of the halo parameters.
We point out that, in reality, however, the rotation curve is rising over
most of the disk in UGC 7321 [242]. Taking this into account affects the
resulting ρ(z) distribution and leads to a slightly puffed up disk ([91]. Hence,
ignoring the rising rotation curve as done by [242] can affect the density
distribution including the HI scale height values; this in turn, will also affect
the halo parameters deduced by [242].
The choice of the gas velocity dispersion value is critical for the result-
ing HI vertical scale height. [242] first choose 9 km s−1 so as to explain
the high observed HI scale height values; and then add a small gradient of
−2kms−1 kpc−1 to improve the fit. Here the resulting vertical scale height
results were fitted with the observed data over the entire radial range (2-12
kpc), since the halo is important from a small radius. For the various gas
parameters tried, the best-fit halo was found to be: isothermal, spherical,
with the central density, ρ0d , in the range of 0.039-0.057 M⊙ pc−3 , and the
core radius, Rc , in the range of 2.5-2.9 kpc. The core radius, Rc , is just
slightly greater than Rd = 2.1 kpc. This is in contrast to the high surface
brightness (HSB) galaxies where the halo core radius is typically a few times
that of the disk scale length, RD [147].
It should be pointed out that the high value of velocity dispersion used,
namely 9 km s−1 ; and even higher values as suggested by the above choice
of velocity gradient, is problematic. This is because the LSB galaxies do
not have internal sources of energy input by supernovae (due to lack of star
formation), or heating by external encounters since LSB galaxies are typically

135
isolated. Both these processes could have otherwise explained these implied
high HI dispersion values required to fit the observed HI scale height data.
Taking account of a rising rotation curve leads to a puffed-up disk, as shown
by [141, 91]. This would imply a higher HI velocity dispersion for pressure
support. This would make the discrepancy between the velocity necessary
for pressure support, and that available in a LSB setting, even harder to
explain.
Thus the self-consistent approach, and the application of two simulta-
neous constraints indicate a dense, compact halo for the LSB galaxy, UGC
7321 [242]. This result by [242] supports and confirms the previous results in
the literature, obtained by fitting the observed rotation curve alone, namely,
that LSB galaxies are dominated by dark matter halo from the innermost
regions (e.g., [243]), and that the stabilization of a superthin disk indicates
a massive, dominant halo [244]. The dense, compact halo deduced by [242]
is shown to explain the superthin nature of this galaxy [147].
We point out that using the same approach of two simultaneous con-
straints except using a non-self-consistent model (Section 6.5), [245] deduce
a spherical-shaped halo for UGC 7321. On the other hand, using the same
data and the same non-self-consistent model as in [245], [34] obtain a prolate-
shaped halo for UGC 7321.
To summarise this subsection: it is surprising that the three galaxies
studied using this technique, and a self-consistent treatment, show a wide
diversity of shapes. We will discuss (in Section 6.5) the various uncertainties
involved in using the HI scale height constraint itself, as one possible cause
of some of this variation.

6.4. Dark matter halo traced using HI scale heights: Non-self-consistent mod-
els in the literature
There are several papers in the literature that have used the HI scale
height as a tracer for dark matter halo; but these papers make ad hoc as-
sumptions, and further they do not obtain a self-consistent vertical distribu-
tion. Thus, the resulting scale heights, or the deduced halo properties, from
such models are not accurate. We discuss some of these here for the sake of
completeness.
For example, the pioneering study on this topic was carried out by [120],
and was applied to NGC 4244 [206] (see Section 6.3). However, there are
several problems with the details of the treatment used: it is not rigorous, and
the vertical distribution of stars and gas is not obtained in a self-consistent

136
fashion. To treat the hydrostatic equilibrium for HI gas, the gravitational
force due to stars and halo are included; however, the effect of gas on stars is
not included. Hence the vertical distribution of stars and gas is not obtained
in a self-consistent fashion as solutions of the equations describing the coupled
system. Instead, an ad hoc choice of a Gaussian is assumed for the vertical
distribution HI gas; and an exponential distribution with a constant scale
height is assumed for the vertical distribution of stars. The shape of the
halo is assumed to be oblate. The equations are further simplified by various
approximations so as to arrive at the shape of the halo. Even in the so-called
multi-component treatment where the stars and gas are treated to be coupled
(see Appendix C in [120]), the equations solved are not general. Rather,
these are simplified and solved analytically using the method by [89], which
assumes the halo to be a perturbation. [89] had developed this approach
to solve for vertical distribution for the multi-component stellar system in
the inner disk, where this assumption is valid. However, it is certainly not
correct to treat the halo as a perturbation in the outer disk region – as done
by [120, 206] – where the vertical force due to the halo dominates over that
due to the disk [19].
Thus, the resulting value of the shape, namely, the oblateness q, obtained
for NGC 4244 by [206] (q = 0.2); and as obtained for NGC 891 (q = 0.2) by
[207], who adopted the method by [120]; are not accurate. The halo deduced
in these two papers is extremely flat.
A similar criticism applies to the studies to determine the halo shape
for UGC 7321 [245], and for a number of galaxies by [34]; who also use an
approach similar to [206], and make a number of ad hoc assumptions to
obtain the scale height. Thus the solution for density distribution obtained
is not self-consistent. These papers preselected a flattened or oblate halo by
choice, and did not even consider prolate-shaped halos – with the exception
of [34] who considered a prolate halo as a possibility for UGC 7321.
Finally, we mention a recent study for a set of face-on galaxies by [246].
Their analysis has many ad hoc assumptions, a few of which are mentioned
here. First of all, unlike the other papers mentioned above in this subsection,
here a purely one-dimensional problem is treated, so that only the pressure
equilibrium of HI is studied, including the vertical force due to stars and the
halo. The halo is assumed to be oblate by choice and obeys a logarithmic po-
tential. The treatment is not self-consistent; instead, an exponential vertical
distribution is assumed for stars and gas, with constant and different scale
heights for stars and gas. Further, a constant gas scale height is assumed;

137
and its value is chosen to be 0.5 kpc or 1 kpc for a particular galaxy. This
assumption of constant gas scale height across the disk is a serious error,
and is in direct conflict with the observed flaring of HI gas seen in many
galaxies. In fact, in such galaxies, the observed flaring of HI is a sensitive
indicator of the halo shape, as we have seen above. Given the various ad hoc
assumptions and errors, the analysis by [246] is unphysical, and the values
of halo flattening as obtained by them are not reliable.
To summarize, the treatment in the papers mentioned in this sub-section
is not rigorous nor self-consistent; hence, the resulting values of the halo
shape, q, obtained are not reliable. Moreover, it is not clear how the result-
ing value of the halo shape, q, is affected given the many assumptions and
approximations made in the treatment. This has been explicitly mentioned
by [34].

6.5. Discussion and Future directions

From the above discussion, it is clear that the few galaxies that have been
studied by this technique; whether using a rigorous approach involving the
self-consistent multi-component disk plus halo model (the Milky Way, M31,
and UGC 7321; see Section 6.3); or, using the non-self-consistent models
(NGC 4244, NGC 891, UGC 7321, ES0 274-G0001, see Section 6.4); both
show a diverse range of halo shapes. That is, the galaxies studied do not
show a preference for a particular shape (also, see [245] for a discussion on
this point). This could represent a genuine variation in shape, although it
difficult to see why halos would form with such a wide range of shapes.
Caveats in using gas scale height as a constraint
Alternatively, the range in halo shapes deduced could also be partly at-
tributed to the shortcomings inherent in the determination of the shape by
this method itself.
The first, and the most serious point is that, the halo extends vertically
much beyond the disk and has a spheroidal (non-disk) distribution. Hence,
one cannot use the effect of its gravitational force on gas at low z to trace
or extrapolate the halo potential or its shape at large z uniquely19 . Thus
the uniqueness of the halo shape deduced this way is not well-established.
In particular, the shape at high z could well be different. An ellipsoidal
shape for the halo, even if it changes with R, has been assumed for simplicity

19
D. Pfenninger 2005, personal communication

138
in the literature; and because this choice is indicated physically for a self-
gravitating system. In reality, the shape may not even be pure ellipsoidal,
especially when taking account of the back-reaction of the disk. This point
has not been noted in the various papers that have used this technique so
far.
Second, the model gas scale height obtained theoretically depends sen-
sitively on the choice of the velocity dispersion and its variation with ra-
dius [16, 78], both these quantities are not known accurately observationally
[16, 78, 34]. This will affect the resulting model values of HI gas flaring. A
small decrease in the gas velocity dispersion, as observed in the Milky Way
(for details, see [78, 19]) – if also applied to other galaxies such as NGC
4244, NGC 891 or M31 – could lower gas pressure support and hence would
indicate slightly less flattened (or more rounded) dark matter halo, as was
pointed out by ([207, 240, 245]; also, see Section 6.3.2). Thus, the results for
q are sensitively dependent on the gas dispersion and its radial variation.
Third, the measurement of HI disk thickness is a challenging task. For
example, even a small deviation in the inclination angle from edge-on (or,
900 ), and the HI optical depth, can affect the values of the thickness measured
as shown in the classic work by [144]. A small deviation in inclination angle
could also affect the rotation velocity measured and this will then affect the
estimated mass of the halo; see [34] for a discussion of this point and other
problems inherent with the technique of using HI scale height as a constraint.
Note, however, that the availability of 3-D HI data cube now overrides these
problems to some degree. This will be further discussed next.
Fourth, here two independent constraints are being used: the rotation
curve to trace the halo potential in the plane, and the HI gas vertical scale
height, to trace the halo potential in the vertical direction. This approach
would, in principle, give an overall better fit to the 3-D density distribution of
the halo by matching the model results with the data, than using either one of
the two criteria singly. However, it is also true that the best-fit parameters;
namely, the core radius and the central density, obtained this way give a
poorer fit to either the rotation curve or vertical scale height, compared to
if the halo parameters were determined by fitting a single constraint alone.
The best-fit parameters obtained using two independent criteria are in a
parameter range that is the overlap from that due each of the constraints; so
that the final best-fit parameter value obtained this way was not the best-
fit obtained using either of the two constraints alone (see the discussion in
[240, 242]).

139
Future Trends:
Despite these caveats, the approach using the HI gas scale height and the
rotation curve still remains the best way to determine the vertical gradient of
the potential and hence the halo shape on a small scale, near the mid-plane
(also; see Section 6.7).
In the past, the main limitation of applicability of this technique has been
the difficulty of measuring the HI scale heights [144]. These measurements
were known for only a handful of galaxies which could be used to study the
dark matter halo using this approach (Section 6.3). This problem may be now
mitigated to a large degree by the availability of modern 3-D HI data cube
for galaxies. There has been tremendous growth in the availability of such
data in the past 10-15 years, since the work reported in the papers in Section
6.3 and Section 6.4. This new data can allow us to map a complete 3-D
distribution of HI including an accurate measurement of the HI scale heights
and rotation curves for external galaxies, without the errors and uncertainties
inherent in the older data and its analysis. (see e.g., [247, 248, 161, 80]).
We urge that it is now time for a systematic study of the halo shape using
this technique for a larger number of galaxies; based on the modern data,
and adopting the self-consistent approach that uses the multi-component
disk plus halo model (see Section 6.3). Such a study would give a systematic
understanding of halo profile and shapes for a larger number of galaxies. It
would tell us if the results show a preference for any particular halo shape; or
if there is any dependence on galaxy type; and if such a result is statistically
significant.

6.6. Shape of halo and its effect on disk dynamics

The shape of the dark matter halo is an important physical parameter.
First, it plays an important role in galaxy dynamics and evolution [249, 250],
and traces the evolution of a galaxy [193, 251]. The shape could depend on
its constituents [195, 252]. Second, it can have important implications for
the disk dynamics, as discussed next. The dark matter halo and its radial
distribution has been well-recognized as being important for the formation
and evolution of a galaxy (see e.g., Section 6.1; also, see [194]).
First of all, it is worth stressing that the dark matter halo has a stabilizing
effect on the disk, which makes it difficult to form planar instabilities [253,
254, 191]. Further, the dark matter halo vertically constrains the disk [16, 19].
This effect is stronger in the outer disk where the halo dominates. Hence the
halo can make the disk resist distortion by a tidal encounter [90]. Although

140
the above dynamical studies considered a spherical halo for simplicity, the
trends in the results obtained would be true irrespective of the halo shape.
The halo shape, and the orientation of the disk in it, are believed to
have a strong impact on the disk dynamics, such as the effect on warps and
bar (see Section 7.2 for a discussion about warps). The origin of warps has
been attributed to: a misaligned angular momenta of the disk and the halo
[255]; to a disk in a tumbling triaxial halo [221]; or a disk that is tilted
w.r.t. the halo mid-plane [256]. A prolate halo shape has been indicated
to explain long-lived warps [257, 249]. However, [231] calculated analytically
the gravitational potential for the progressively more prolate halo as deduced
for the Milky Way by [140]; and showed that the shape of the halo has little
effect on the longevity of the warp. Similarly, the disk response, and hence
the radius of onset of warps, is shown to be nearly independent of the shape
of the halo [258]. On the other hand, the bar evolution is shown to be affected
by the shape of the halo [259]. Thus the role, or even the possible importance,
of halo shape on disk dynamics is not well-understood; this topic needs to
be explored more.

6.7. Other tracers and complementarity

Here we briefly discuss the other tracers that have been used to measure
the halo shape, so as to give a broad coverage of the topic. It is difficult to
study the halo shape, in particular the shape normal to the galactic plane;
since one needs sufficient number of luminous tracers, that are distributed
normal to the galactic plane, up to a large distance from the mid-plane. In
the literature, various tracers have been used to study the halo shape. For
example, stellar streams in the Milky Way have been used as a tracer; and
these indicate the halo to be oblate, or such studies have used an oblate halo
as a preselected choice while analyzing the data [260, 261, 262, 263]. On
the other hand, some studies using stellar streams indicate a prolate halo
[264, 220]. For a recent review of stellar streams as a tracer of dark matter
halo, in the context of Gaia, see [265]. Globular cluster distribution in the
Galaxy indicates a spherical halo in the central 20 kpc [266].
There are, however, some cases where a prolate-shaped halo is strongly
indicated, as in polar ring galaxies [267, 268]. In some cases, the halo shape
can vary with radius within a single galaxy, as for the Milky Way ([140, 180];
see Section 6.3.1); and in external galaxies [268].
The shape of the dark matter halo measured by different tracers seems
to depend on the choice of the tracer used, and it may also be sensitive to

141
the physical region where the tracer is located. It could also depend on the
properties of the constituents of the dark matter halo [195, 252]. Due to
the limitation of resolution of data to be used as a constraint, typically a
constant shape across the entire halo is used, for simplicity. The paucity of
some of these tracers (e.g. globular clusters) means that using these tracers,
the potential cannot be sampled uniformly. Hence, these are less reliable as
tracers of the halo shape than, say, the HI vertical scale height as a tracer.
For a good summary of different tracers used, and the resulting shape of the
halo deduced, see Section 6 in [220].
Most of these other tracers are located at a large distance from the mid-
plane; hence they trace the halo potential at high z values. This is comple-
mentary to the information on the halo shape deduced from HI distribution,
which is distributed in a thin disk close to the mid-plane.
In summary, the halo shape as deduced from various tracers seems to
show a bewildering variety of shapes and a range of axes-ratio values, even
within large spiral galaxies. The physical origin for this diversity has not
been clearly understood or discussed much in the literature. This needs to
be addressed. We suggest that the variety of halo shapes deduced reflects to
some degree the physical region occupied by the tracer and/or the uncertainty
with the technique using a particular tracer. The last point was discussed in
Section 6.5, to show that this could be partly responsible for the variety of
halo shapes seen in different galaxies using the same tracer, namely the HI
vertical scale height.
To conclude, a systematic and comprehensive study of the halo profiles
and shapes using the HI scale heights as well as other tracers is required, to
get a full understanding of the halo parameters.

6.8. MOND (MOdified Newtonian Dynamics) as an alternative to Dark mat-

ter halo
In this review, we have taken the standard λCDM approach; where the
dark matter halo has been evoked to mainly explain the nearly flat rotation
curves observed in the outer regions of spiral galaxies (Section 6.1). It is
interesting to ask whether at low accelerations as seen at large R and large z
away from the mid-plane, an alternative theory of gravity, namely, MOdified
Newtonian Dynamics (MOND), could be relevant. MOND, an alternative to
Newtonian gravity, was proposed by [269]; meant to be applicable in regions
of low accelerations, such as the outer disks of galaxies. This was meant
to explain the observed, nearly flat rotation curves without the necessity of

142
evoking a dark matter halo. Since then much work has been done on the
various forms, and tests, of MOND (see, e.g., [270, 271, 272]). The effect
of MOND on the flaring of HI gas in the outer disk of the Milky Way, was
studied by [273]. They showed that the observed HI flaring between 17-40
kpc could be explained reasonably well by MOND; but in the range of 10-16
kpc, the HI data showed deviations from the prediction of MOND.
We note, however, that stars constitute the main mass component of the
disk. Hence it would be worth checking whether MOND can explain the
scale height values, including the flaring of the stellar distribution that is
seen within the optical disk of Milky Way and other galaxies (see [16, 35];
and Sections 4, 5). It has been shown (see [16]; or, Fig. 6) that for the
observed parameters, under Newtonian dynamics, a stars-alone disk in hy-
drostatic equilibrium shows exponential flaring. The inclusion of gas gravity
in the coupled system brings down the stellar scale heights which match well
with the observed values for R < 12 kpc. In this radial range, the dark
matter halo contribution to gravity is negligible, and it does not affect the
scale heights of stellar distribution significantly [16]. MOND gives an excess
effective gravitational force compared to |Kz |, the Newtonian vertical force
per unit mass (e.g., [272]). Since Newtonian gravity already gives a good
agreement with observed scale heights within R< 12 kpc [16]; therefore, on
the face of it, applying MOND to this radial region would give a mismatch
between predictions from MOND theory and the observed data for stellar
scale heights. Hence, MOND does not appear to be favoured to explain the
observed stellar scale heights in the inner Galaxy (R < 12 kpc). However,
this should be checked more quantitatively to confirm whether MOND can
be ruled out from the study of disk vertical structure; or, whether it does play
some role in the determination of the stellar vertical scale heights, including
flaring, in the Milky Way stellar disk.
As discussed above, MOND predicts a higher |Kz | than the Newtonian
value. It has been shown that the MOND effects are expected to be significant
when the disk surface density falls below the critical surface density of 137
M⊙ pc−2 [272]. We show that in the the Milky Way, with central surface
density of 649 M⊙ pc−2 and RD = 3.2 (see Section 4.2.3); the above critical
surface density will occur at R∼ 5 kpc. Hence, the MOND effects should
become apparent at radii beyond R=5 kpc. Thus the radial range of 5-
12 kpc is the ideal radial range where the future tests for MOND versus
Newtonian dynamics should be carried out. Other possible indicators of the
vertical dynamics, such as the tilt of the velocity ellipsoid, do not yet clearly

143
favour either λCDM, or MOND, as discussed by [272].

7. Related topics
The focus of this review is on the equilibrium vertical structure of a
typical ”thin” galactic disk. In this section, for the sake of completeness,
we briefly discuss a few other important topics related to the study of disk
vertical structure. These include the kinematically and chemically distinct
Thick disk component (see Section 1 for the definition) in the Milky Way;
and various dynamical features and instabilities along the vertical direction.
These are well-developed and active topics of research in their own right. For
the sake of keeping the scope of this review within limits, we do not cover the
above topics in detail. Instead, here we briefly mention the salient points, and
unresolved questions about these related topics, and point to a few important
references. The material covered in the earlier Sections (Section 4.2, Section
4.3, Section 4.4) has direct relevance for some of these topics, in particular
Section 7.1, as will be pointed out in this section.

7.1. Thick disk

The Milky Way, as well as other galaxies, contain a kinematically and
chemically distinct, metal-poor Thick disk component; in addition to the
thin disk component which we have focused on in this review. The topic of
Thick disks is a vast, well-developed topic that is evolving rapidly, triggered
by new data and studies involving numerical simulations. The aim of this
subsection is to give a brief glimpse into this topic.
Starting from the work of [274] in 1980’s, over the past forty years, it has
been clearly established that the Milky Way has a Thick disk, which is thicker
than the canonical thin disk. However, its properties including the thickness
and the fraction of mass in it compared to the thin disk, or its origin, are yet
not well-understood [8, 9, 275]. The surface density distribution of the thick
disk falls exponentially with radius. The typical scale length for this fall-off
is measured to be ∼ 2 kpc [276, 277] vs. the typical scale length of ∼ 3 kpc
for the thin disk (e.g., [2, 82]). Thus, most of the mass in the Thick disk is
in the inner Galaxy, well inside R=8.5 kpc.
As the name suggests, the Thick disk component is vertically more ex-
tended and kinematically hotter than the thin disk component. The thickness
of the vertical density distribution of the Thick disk is in the range of 0.5-1.2

144
kpc [8] – for example, a value of 750 pc is obtained using SEGUE photome-
try [278]; versus a thickness of 300 pc determined for the thin disk [2]. The
Thick disk mass is now believed to be comparable to that of the thin disk
[160, 279, 280, 281, 275]. The rotational velocity of the Thick disk lags w.r.t.
the thin disk by ∼ 50 − 100 km s −1 [282, 283, 284, 285].
The Milky Way Thick disk may, in fact, contain several sub-components
[277]. Also, there could be structural continuity between the thin disk and
the Thick disk [286, 277]. Thus dividing a galactic disk into two discrete
components – a thin disk and a Thick disk – may be an oversimplification
[286, 287]. However, for the sake of simplicity, most studies still use two
discrete components: namely, a thin disk and a Thick disk (see the discussion
in [287]).
A Thick disk is common in external galaxies as well [288, 159, 160, 182,
289, 290, 291]. Thick disks were first discovered in external galaxies as an
excess of light over the thin disk in edge-on galaxies ([292, 26, 160]). Typi-
cally, a sech2 function is used for the density profile to analyze data to study
Thick disks [159, 293]. However, this choice of density may not be correct
for a real disk. As already shown by [160], a correct, physically motivated
multi-component treatment and the resulting self-consistent density profiles
– which are different from sech2 – affect the deduced values of mass contained
in the thin and Thick disks (see Section 4.5.1). [160] concluded these to be
comparable. They argued that this favours the scenario of in situ formation
of most of the stars in the Thick disk.
The above important study by [160] needs to be updated using the thick
disk approach (as in Section 4.3), and the inclusion of the effect of the halo,
as discussed next (also, see Section 8). For the study of vertical structure of a
Thick disk, some of the topics developed in this review would be particularly
relevant: such as, the treatment for a physically thick disk where the R
term in the disk Poisson equation needs to be kept in (Section 4.3); the
kinematical effects incorporated through the complete Jeans equations; and
the inclusion of effect of non-isothermal velocity dispersion (Section 4.4). To
model a Thick disk properly, a correct, multi-component treatment needs to
be done, that includes the thick disk approach (as in Section 4.3), and the
kinematical effects and non-isothermal velocity dispersion (as in Section 4.4),
as well as the effect of the dark matter halo. The vertical structure of Thick
disks needs to be studied systematically.
The origin of the Thick disk is not yet well-understood. Various scenar-
ios have been proposed for the formation of the Thick disk. These include

145
Figure 30: A schematic diagram of bending waves and breathing modes, which illustrates
the different bulk motions that characterize these two types of features – see the text for
details. Source: S. Ghosh

internal mechanisms such as heating of the thin disk by spiral arms; radial
migration within the disk; or external mechanisms such as accretion of stars
during a merger with a satellite galaxy [8, 127, 275]. However, given the
observed structural continuity between the two disks, the actual origin of the
Thick disk may be more complicated; and may involve both internal as well
as external mechanisms [275].
The topic of Thick disks in galaxies remains extremely active in all as-
pects: namely, its structure, composition, dynamics and evolution.

7.2. Dynamical features along the vertical direction

We have discussed a steady-state distribution in a galactic disk in this
review. In addition, the disk also exhibits many complex dynamical fea-
tures along the vertical direction: such as, warps and corrugations (both
attributed to bending waves); and breathing modes – see Sections 7.2.1 and
7.2.2, respectively, for details. Both these types of features involve coherent
oscillations of stellar disk in the direction perpendicular to the mid-plane
[294, 295, 296]. These features are collective excitations [297, 298, 295]. The

146
associated vertical motions are of two distinct types. The one in which stars
on either side of the mid-plane move coherently in the same direction normal
to the mid-plane are called bending waves; while the other, in which stars on
either side move coherently in opposite directions, namely either compressing
towards or expanding away from the mid-plane, are called breathing modes.
See Fig. 30 for a schematic diagram which illustrates the different bulk mo-
tions shown by these two types of features.
Recent evidence for both these types of features in the Milky Way comes
from large-scale, non-zero bulk vertical motions (∼ 10 km s−1 magnitude):
as seen from Gaia ([299], see their Fig. 6 c; also, see [300, 301, 302]; as well
as evidence obtained from previous Galactic surveys [294, 303, 304, 305]. For
an axisymmetric disk in a steady-state, the bulk motions along the radial
and vertical directions, vR and vz , respectively, are expected to be zero (e.g.,
[2, 296]). Hence, it is important to understand the non-zero bulk motions that
are observed and their dynamical implications. The number density and bulk
velocity of stars in the solar neighbourhood show a North-South asymmetry;
which indicates a wavelike pattern, probably triggered by a satellite encounter
[294, 306].
These are complex, active topics and much work has been done on these
already. For example, warps have been long-studied as features that indicate
an m=1 or cosϕ dependence of density along the z direction. Early evidence
for bending waves such as warps in the Milky Way, and the evidence in other
galaxies even now, mainly involves information about the shape or the spatial
distribution of the tracers [307, 308, 309]. Only in recent years, it has become
possible to measure the bulk motions associated with warps in the Galaxy,
as discussed above.
In addition to the above two types of features, recent observations from
Gaia have shown evidence for the existence of a novel, and unexpected,
phase-space spiral. This is an extremely active topic of research now (Section
7.2.3).

7.2.1. Bending instabilities leading to warps, and corrugations

A typical galactic disk exhibits a variety of vertical features w.r.t. the
uniform disk distribution, such as warps and corrugations, which can be
described in terms of bending instabilities. An axisymmetric disk under an
imposed or external non-axisymmetric potential can produce bending waves
w.r.t. the mid-plane (z=0) such that at any location (R, ϕ) in the disk, the
vertical displacement ∆z ∝ (cos mϕ), where m is the azimuthal wavenumber

147
Figure 31: A striking example of a stellar warp in the galaxy ESO 510-G13. The image
shows a distribution of stars along z (including the dust lane at the mid-plane) that clearly
follows an m = 1 azimuthal variation, indicative of a warp in an edge-on galaxy. Source:
NASA/Hubble Heritage Team

(see [310] for a clear discussion of this topic, and the references therein). If
the mode is seen across the entire azimuthal range at a given radius, R, then
the wavelength of the bending waves satisfies the relation mλ = 2πR. Here
m = 1 denotes warps; these are commonly observed in galaxies. Higher order
modes (m > 1) have smaller amplitudes and are seen as corrugations of the
plane. The study of bending instabilities has a long history, despite which
the origin and dynamics of bending waves is not fully understood. Hence the
topic is still very active.
A variety of physical mechanisms have been proposed in the literature
for the generation of bending waves. The idea that warps are free modes of
oscillations was first proposed by [311]. The most studied mechanism for the
origin of warps is a tidal encounter with another galaxy [312, 313, 314, 315].
When triggered by a tidal encounter, if the velocity of the satellite normal
to the disk is smaller than that of the stars, the perturbation is taken to be
mainly a bending mode [295]. An encounter with a satellite galaxy can cause
vertical oscillations, corrugations and flaring of the stellar disk, as shown

148
 Z (arcsec)
Z (arcsec)

20
10
0
-10
-20

20
10
0
-10
-20
-100

-50
-50 0
X (arcsec)

0
X (arcsec)
50

50
100




!"#$%&'
Figure 32: Dust corrugations in NGC 4302: The top left panel shows the Spitzer/IRAC µ
image; bottom left is the SDSS R-band image given after unsharp masking. The middle
panel shows the stellar mid-plane (solid line) and the dust mid-plane (blue filled circles),
derived from the left-hand panel images: this clearly shows the dust corrugations. The
right-hand panel gives the result of Fourier analysis on the dust mid-plane. The mean
wavelength for the two peaks in the power spectrum corresponds to a corrugation of size
∼ 3 kpc, where 1 arcsec ∼ 76 pc. Source: Taken from [310]

in the simulations study by [316]. The frequency of strong warps has been
shown to be higher at a high redshift (up to z=2) by [317] from a study
based on the HST and JWST observations. They claim that this is due to a
higher frequency of interactions and mergers at high redshift. The bending
instabilities could also be generated by internal processes such as instabilities
[318, 319]; or due to interaction with a bar [320]. See [310] for a discussion
of other external and internal dynamical processes for the origin of bending
instabilities.
Warps have been the most studied of the bending instabilities: as seen
observationally in HI [307, 308, 321, 322]; and in stars [309, 323, 324, 325],
both in external galaxies. See Fig. 31 for a striking example of a stellar warp
in the galaxy ESO 510-G13. A Warp have been observed in the Milky Way,
using HI as a tracer [42, 111, 110]; and using stars as a tracer ([256], also see
references for the stellar warp in the Milky Way in Section 2.1.1). The warps
have also been extensively studied theoretically [312, 326, 327]. Despite this,
however, the origin and dynamics of warps is not yet well-understood.
At the other end of the size spectrum, the higher order (large m, or small
size and small amplitude) perturbations, known as corrugations, are harder to

149
observe; and hence have not been so well-studied. The higher order bending
waves with m ∼ 10 or larger, known as scalloping, have been observed in HI
at large R in the Milky Way [328, 329, 111].
Corrugations have been observed in stars in a few external galaxies: in
NGC 4244 and NGC 5023 [330], and in NGC 5907 [331]. Recently these have
been observed even in dust in a sample of external galaxies [310], see Fig.
32 for the corrugations in dust seen in NGC 4302. Corrugations have been
detected in an edge-on superthin LSB galaxy, IC 2233, both, in stars ([332]),
and in HI gas ([333]).
A tidal encounter has been proposed and shown to be the physical mech-
anism for the origin of corrugations in stars and gas in a galactic disk; see
the recent simulations study by [334]; also, see [316].

7.2.2. Breathing motions normal to the galactic disk

The Galactic stellar disk also shows non-zero vertical bulk motions called
breathing modes, with properties as described above; namely, where stars
on both sides of the Galactic mid-plane move coherently towards or away
from it, see: ([294, 295]; [299] - see Fig. 4; [300, 301, 302]; and [296, 335],
and references therein). Of the above references, [294, 295] are based on the
SDSS data; while the rest are mainly based on the detailed kinematic data
from the Gaia Data Release 2 (Gaia DR2) – except for [335], which is based
on Gaia DR3 data.
The breathing modes could arise directly due to a tidal encounter with
a satellite or a galactic sub-halo [295, 336]; or they could arise due to an
internal cause such as spiral density waves [337, 338, 339], or, a bar [340]. By
studying a self-consistent, high-resolution simulation which has prominent
spiral arms, [296] have shown that breathing motions are induced by the
spiral arms – see Fig. 33. This figure shows the signature of the breathing
motions as proposed from their model, namely an increase in the vertical
velocity, vz , with z associated with a spiral arm. They show that this result
is confirmed by the Gaia DR2 data. Further, in a follow-up work, [341] show
that the breathing motion is caused by the spiral features generated in the
simulation of a fly-by encounter of a satellite. They show that the cause
is not the encounter itself as has been argued in the past (e.g., [295, 336]);
rather, it is due to the spiral arm triggered by the encounter. This point has
been confirmed by the recent simulations results by [342].
There is observational evidence – from analysis of Gaia DR3 data – for
compressing or expanding breathing motions being aligned w.r.t. various

150
Figure 33: Breathing motions: Plot of mean velocity < vz > vs. z of stars (between
the given x-y range) responding to the spiral pattern in the simulation. The red and
blue lines represent the response of the older and younger stars, respectively. This plot
clearly shows the breathing modes as predicted by [296]: a positive slope would indicate an
expanding (positive) breathing motion, while a negative slope (as in this figure) indicates a
compressing (negative) breathing motion. The younger stars show a stronger effect. This
result is shown to agree with a similar plot obtained using the Gaia data for the Milky
Way (see Fig. 14 in [296].) Source: Taken from [296]

spiral arms in the Galaxy [306, 335]. These studies show that the Local arm
and the Outer arm are associated with compressing breathing motions, while
the Perseus arm is associated with an expanding breathing motion. In an
interesting study, [335] show by simulations of an isolated Galaxy, that a
transient spiral arm during its growth phase induces compressing breathing
motion in the disk stars, while an expanding breathing motion occurs when
the arm is getting disrupted. Hence they infer that the Local arm and the
Outer arm in the Galaxy are in the growth phase, while the Perseus arm is
in the disruption phase. Therefore, [335] suggest a general interesting result
that the sense of breathing motion (compressing vs. expanding) can be used
as a proxy to decide whether the spiral arm that gave rise to the breathing
motion is in a growth phase or a disruption phase, respectively. The topic of

151
Figure 34: Phase space spiral in the Milky Way [343]: Plot of distribution of stars in the
vertical position-velocity (z − vz ) phase space for stars sampled within the galactocentric
radial range of 8.24-8.44 kpc; coloured as a function of median azimuthal velocity, vϕ , in
bins of ∆z = 0.02 kpc and ∆vz = 1 km s−1 . vϕ is taken to be positive along the direction
of Galactic rotation. Source: Taken from [343]

breathing motions is emerging as an active area of research now.

7.2.3. Gaia Phase-space spiral

One of the most enigmatic discoveries from the Gaia DR2 data [299] is
the existence of a prominent spiral feature in the vertical, or z − vz phase
space of the the distribution function of stars, in the solar neighbourhood
[343], see Fig. 34. Here vz is the velocity normal to the disk mid-plane.
This has come to be called the Gaia phase-space spiral or the Gaia phase-
space snail in the literature [299, 343, 344, 345]. This phase-space feature
has a fractional amplitude of a few percent and a complex morphology that
depends on the galactic radius, azimuthal location, and angular momentum
[316, 346, 347, 348, 344]. The phase-space feature has a spatial scale of ∼ 200
pc and a velocity scale of ∼ 10 kms−1 [344]. A phase-space spiral has also
been detected in Gaia DR3 data [349]. For a recent review of this topic, and
the Milky Way dynamics in general, in the light of Gaia data, see [11].
The phase-space spiral is believed to be a disturbance in the z − vz dis-

152
tribution function from a past event, such as a tidal encounter, that has
undergone phase mixing [350, 346, 316]. This idea of a kinematical origin
leads to an estimated age of ∼ 300-900 Myr for the phase-space spiral [343].
The Galactic vertical potential near the mid-plane can be represented as an
anharmonic oscillator; hence, the orbits with a smaller z amplitude have a
higher vertical oscillation frequency (see Eq. 2 in [343]). Therefore, in this
picture, a perturbation gets sheared into a vertical spiral that gets more
tightly wound with time (e.g., [351]). Thus, the phase space spiral as a kine-
matic feature is shown to get wound up with time. [343] point out that
this estimated age agrees well with the epoch of perturbation due to the last
peri-centric passage of the Sagittarius dwarf galaxy, as seen in simulations;
which thus could have triggered the phase-space spiral.
The origin and dynamics of the phase-space spiral is an extremely active
topic of research now. The existence of the phase-space spiral, and the asso-
ciated observed non-zero bulk motions along radial and vertical directions,
vR and vz , respectively; are taken to be an evidence that the disk is out-of-
equilibrium ([343, 352], and references therein). This is because a stationary
and axisymmetric disk will have zero bulk motions along R and z direc-
tions. Alternatively, the existence of phase-space spiral is taken to indicate
that the disk is only approximately in a steady-state [350, 344]; or is under
a time-dependent potential [187] where Jeans equation is no longer valid.
During simulation studies of a tidal encounter, [187] find that in low density
regions, the estimate of surface density is not given correctly by the usual
Jeans equation. Rather, it is an overestimate, which they propose points to a
time-dependent potential. A tidal encounter with a satellite galaxy has been
proposed as one possible physical mechanism for the origin of the phase-space
spiral [343, 316, 346, 187, 353]. [336] propose that a tidal encounter gives rise
to both bending and breathing modes in a galactic disk, which then undergo
vertical phase mixing which gives rise to local phase-space spirals.
It has now been realized that a kinematic phase-space spiral generated by
a single event is not adequate to explain all its properties (e.g., [344], and the
references therein). For example, N-body simulations have shown [354] that
an encounter with the Sagittarius dwarf (often applied in this context) cannot
match the observed amplitude of the phase-space spiral in the Milky Way.
Another possibility [345] is that the phase-space spirals could repeatedly arise
due to a series of perturbations in potential, say, due to galactic sub-halos.
These features would then get dissipated due to encounters with the giant
molecular clouds in the disk [345]. This would lead to an apparent age or the

153
net duration of the phase-space spiral to be ∼ 0.5 Gyr. Thus, irrespective of
the physical mode of origin of the phase-space spiral, it would get erased by
scattering off the giant molecular clouds or other small-scale features within
< 1 Gyr. Thus, the existence of a phase-space spiral may not have an origin
due to an encounter, and the kinematic argument for its age would then not
be applicable.
Recently, it has been argued [344] that the phase-space spirals could be
swing-amplified perturbations [355, 356] in 3-D, where the self-gravity plays
a crucial role in the evolution of the features. Further, a co-rotating cloud
could excite stationary phase-space spirals. Hence, a simple argument, based
on the epoch of tidal encounter that is used to estimate its age, may not
be applicable [344]. A wake generated in the dark matter halo due to an
encounter with a satellite can also give rise to a long-lived phase-space spiral
along z [357].
A phase-space spiral, indicating a deviation from equilibrium, has now
also been detected from the Gaia DR2 radial velocity catalogue [358], by
analyzing the residuals of an equilibrium model. Ripples in the plane have
been known for some time [359]; also, a phase-space spiral in the R-vR phase
space has now been discovered [360, 361]. This is a rich and an extremely
active subject now. An introduction to the literature and the current ideas
on the topic of phase-space spirals is given in this section, citing the relevant
papers; so that the reader may get an idea about this fast-developing topic.
Phase-space spiral: An out-of-equilibrium feature
In summary, the phase-space spiral features are intriguing dynamical enti-
ties. They are often taken to be an evidence of a disk that is not in a steady-
state or is out-of-equilibrium (e.g. [343]; also, see the discussion above).
However, we emphasise that that the amplitude of the spiral features is only
a few % of the background; so the entire disk, or the entire 6-D phase space
of the disk, may not be out-of-equilibrium. After all, most disks seem to sat-
isfy a density profile as obtained from an equilibrium model (Section 4.2.6;
Appendix A). The associated results from such equilibrium models, such as
the disk scale heights for stars and gas agree well with observations for the
Milky Way [16].
Moreover, the origin and life-time of the phase-space spirals is not yet
well-understood. These could well be repetitive or generated often; or could
be long-lived -as when generated by a co-rotating cloud, as discussed above.
Hence, perhaps it is more appropriate to call the phase-space spiral to be an
out-of-equilibrium feature, rather than it being an evidence for the entire disk

154
being out-of-equilibrium. This distinction is important for understanding the
overall vertical structure and dynamics of the disk. As an interesting example
in support of this argument, consider the study by [362] who do the analysis
of the phase-space spiral, by treating it as a perturbation on the underlying
disk potential; and show that this gives information on the galactic potential
that is complementary to that given by the bulk of the (undisturbed) matter.
Finally, even if the phase-space spiral denotes an out-of-equilibrium fea-
ture, so that the disk does not strictly satisfy the vertical Jeans equation; the
steady state analysis of a disk in vertical equilibrium as given in this review
will provide a useful background to understand the dynamics of the phase-
space spiral. It will give the necessary groundwork or the starting point on
which to build the dynamical picture of a non-equilibrium feature such as a
phase-space spiral. We would like to point out that this is analogous to the
treatment of a warp or other bending instabilities like corrugations, which are
also treated as perturbations on the basic undisturbed, axisymmetric disk;
this approach is routinely taken to study these features (see Section 7.2.1). .
Indeed, [358] have treated the phase-space spiral as a perturbation in an
undisturbed disk. In fact, the dynamics of phase-space spiral has now been
used to identify and understand an out-of-equilibrium local region in a disk
that is overall in equilibrium [358, 352]. This is a promising line of research
in this topic.

8. Vertical structure and multi-component disk plus halo model:

Future directions
The field of vertical structure of galactic disks promises to remain exciting
and active; driven mainly by new observational data - especially kinematic
data, and data for stellar number density; and numerical simulations. A
few important future problems in this field, in particular those motivated by
the multi-component disk plus halo model, are briefly described below. The
model is timely to explain the new data from Gaia DR3 and DR4 for the
Milky Way, and the data from JWST for high-redshift galaxies.
Direct Results and Applications from the Model:
First, more accurate input parameters for stars including kinematical
data are now available for the Milky Way, for example, from Gaia. For ex-
ternal galaxies, the various IFU surveys are beginning to provide information
on the 2-D kinematics (see Section 2.1.4), including the stellar velocity dis-
persion. With better information on the input parameters, more accurate

155
model results for the self-consistent vertical stellar density distribution, and
the associated scale heights, can be obtained even for external galaxies. This
could, for example, further check and confirm whether a flaring stellar disk
is a common feature of galaxies (see Section 5). For the Milky Way, the
theoretical model density profiles can be compared with star counts data for
example, from Gaia, and LAMOST.
However, a practical point of concern is as follows. Due to resolution
problems, the stellar velocity dispersion measurements from many of the
IFU surveys are still limited to the inner regions only. For modeling the
density distribution, one would like to know the stellar velocity dispersion
as a function of radius in the disk. This requires sufficiently broad angular
coverage; and also a small sampling distance, or, good resolution. Both these
conditions are well-satisfied, for example, in the CALIFA survey, where the
measurements cover a distance of ∼ 1 − 2Re (see the discussion in [60]). The
stellar velocity dispersion vs. radius thus measured is given in Fig. 7 in the
detailed work by [60]; or, see Fig. 35. In other surveys, often a single average
value for the dispersion, say within the effective radius, is extracted from the
data, as done from the SAMI data by [363]. A detailed analysis on lines of
[60] needs to be done for data from other surveys to extract the values of
stellar velocity dispersion as a function of radius.
Another future problem worth doing is as follows. With the detailed
information about the stars by the subtype in the Milky Way now available
from Gaia, it would be possible and worthwhile to formulate a model with a
multi-component stellar disk (that is, a disk consisting of various sub-types of
stars), and which also includes the effect of gas and halo (see Section 4.2.1).
Second, the multi-component disk plus halo model and analysis (Sec-
tions 4.2 to 4.4) can be naturally extended to study the vertical structure
of the kinematically and chemically distinct Thick disk in the Milky Way in
a systematic way, as discussed in Section 7.1. To do this, a correct multi-
component disk plus halo treatment that includes a thin disk, a Thick disk,
treated in the field of the halo, would need to be formulated in the (physi-
cally) thick disk approach (Section 4.3). If the input parameters for the thin
disk and a Thick disk are known from observations, as in the Milky Way (see
Section 7.1); then the above procedure can give the density profile, ρ(z), for
each disk component.
On the other hand, when the disk parameters are not known well – as
in external galaxies – then matching the above resulting model profiles with
the observed intensity profiles, the parameters of the thin disk and the Thick

156
Figure 35: CALIFA Survey results for stellar velocity dispersion in galaxies [60], given
in units of km s−1 . The rows from left to right denote the SDSS images of galaxies, the
ATLAS3D velocity dispersion map, the CALIFA velocity dispersion map, and the radial
profile for the velocity dispersion (in colour for CALIFA and in grey for ATLAS3D . The
size of the ellipse shows one effective radius and is indicated by the dashed line in the
right-most panel. Thus, the CALIFA survey typically measures values of stellar velocity
dispersion beyond one effective radius, Re . Source: Taken from [60]

157
disk can be constrained, simultaneously. This would be on lines of what was
done by [160] except that they had used the simpler thin disk approach; and
they did not include the effect of the dark matter halo (see Section 4.5.1,
and Section 7.1). Thus, the approach outlined here would be a more realis-
tic treatment of the problem. Since Thick disks are ubiquitous in external
galaxies (see e.g. Section 7.1), this study would provide an insight into the
vertical structure of Thick disks in these galaxies.
Third, the multi-component disk plus halo model can now be applied to
study the vertical structure of the new regime of high-redshift galaxies. This
exciting area of study has opened up in the last couple of years with the ad-
vent of data from JWST (James Webb Space Telescope), along with ALMA
(Atacama Large Millimeter Array). The latter provides information about
the gas content of galaxies. The data from JWST, and ALMA, have revealed
surprising new features about the high-redshift galaxies. These galaxies are
found to form early on; are more massive [364, 365, 366]; and gas-rich [367]
compared to the present-day galaxies; and are more evolved than had been
expected. These are presumably the precursors of present-day galaxies like
the Milky Way. Hence it is important to understand their vertical structure,
which could contain evidence about galaxy evolution (see Section 1).
Using the data for a large sample of 181 galaxies between a redshift of
z=0 to 5, [368] have obtained the disk thickness for this sample. This is
done by fitting the JWST NIRCAM imaging observational data with a single
disk with a sech2 distribution function. [368] find that at z > 1.5, the disk
thickness is comparable to that of the Thick disk – and is much larger than
the thickness of the thin disk of the Milky Way. They claim that this points to
the formation of a Thick disk at early times. A more detailed study involving
a 3-D modelling of the imaging data from CEERS, JADES, and PRIMER
surveys using JWST has been done by [170], where they do a double disk
fitting – each with a sech2 function – for a sample of galaxies. They show
that a second disk is necessary to explain the observations.
However, both these studies have assumed a sech2 profile – which is only
applicable for a one-component, isothermal disk. Recall that the stellar
density profile in a real galaxy is typically steeper than sech2 , as seen in
observations (Section 2); and also as obtained theoretically for a realistic,
multi-component disk plus halo case (Sections 4.2.4 to 4.2.6, Appendix A).
Therefore, we suggest the following better and physically more correct ap-
proach: namely, obtain physically motivated density profiles by applying the
multi-component disk plus halo model to a coupled thin and a Thick disk

158
system; and compare the model results with the JWST data. This will be
on lines of the application by [160] to the S4 G data; except that, here the
more general model involving the thick disk approach (Section 4.3), and other
kinematic effects (Section 4.4), plus the effect of halo, would be included to
make the treatment more realistic and complete. This procedure was out-
lined above in the context of the second future problem given earlier in this
Section. By comparing the model profiles with the JWST data, one can
then extract the physical parameters of profiles of the two disks: an impor-
tant future problem. This could also help one understand the dynamics and
evolution of high redshift galaxies.
Dynamical implications
In addition to the above problems involving direct application of the
model, there are important implications of the model results for other dy-
namical problems. A few of these are outlined below.
First, the observed HI scale height values and rotation curves were used
as two independent constraints on the multi-component disk plus halo model,
to get the best-fit values for the shape and density profile of the dark matter
halo (Section 6). The 3-D HI data cube for galaxies now available allows
an accurate determination of the input parameters: such as the HI gas scale
height values (which were not easy to observe so far), and an accurate rotation
velocity measurement as a function of radius ([247]; also, Section 6.5). Hence
the above approach can now be applied systematically to a large number of
galaxies to obtain their halo properties.
Second, due to the vertical constraining of a disk component in a coupled
system ([16, 18, 19]; also, Section 4.2.5); the disk is more tightly bound
closer to the mid-plane. That is, it takes more energy to move a unit mass
by a certain height [90]. Therefore, a coupled disk is predicted to be more
resistant to distortion; say, due to a tidal encounter. Detailed numerical
simulations are needed to check if a realistic, multi-component disk is indeed
more robust to distortion, as predicted by [90]. This effect is expected to be
higher in high redshift galaxies since these are gas-rich, as seen above. This
could have important implications for the early evolution of galaxies.
Third, the model would have important implications for other dynam-
ical problems in the field, such as warps and other bending instabilities.
Typically, in dynamical studies of such features, the disk vertical profile is
assumed to be of sech2 form – as applicable for a single-component, isother-
mal disk (e.g., [294, 295, 336]); or, to have other simple analytical form

159
[369]. An early work that explored the dependence of results on the density
profiles [258] showed that realistic vertical stellar density profiles, which are
less steep than an exponential, allow warps to set in from a smaller radius,
namely, 2-4 RD – which lies within the optical disk size (see Appendix A).
This explains why optical or stellar warps are ubiquitous (see Section 7.2).
The steeper-than-sech2 profile and a vertically constrained distribution ob-
tained in a multi-component disk plus halo model, and as is seen in observed
data; and the associated change in the gravitational potential and force; are
likely to affect the detailed properties of warps and other bending modes.
This general idea needs to be investigated further.
A related point is as follows. In studies of galactic dynamics, typically a
simplified analytical form for the potential is used, for example the popular
Miyamoto-Nagai potential – for its mathematical ease of use [370, 2]. How-
ever, while this does give the correct radial profiles of surface density, it does
not represent the vertical profiles such as sech2 or an exponential [371]. It
has now been realized that the detailed vertical density profile could play an
important role in galactic dynamics. Hence a form for potential-density pairs
that represent vertical density profiles such as sech2 or an exponential has
been recently obtained in an important paper by [371]. This would be useful
to study dynamics of disks; for example, for the evolution of a phase-space
spiral where the anharmonic term in the potential is important in deciding
the winding time [371, 343].20

9. Conclusions
In this review, we have tried to give a cohesive, physical introduction
to the vertical structure and related dynamics of a typical, thin galactic
disk in hydrostatic equilibrium, while keeping in mind recent observations.
The treatment is comprehensive, starting from the classic, one-component,
isothermal disk [12]; up to the modern picture involving a more realistic,
multi-component disk plus halo model [16, 19]. This model includes both
stars and interstellar gas in a galactic disk on an equal footing. These are

20
To put this in the context of the current review, we emphasize that the self-consistent
density profiles resulting from a realistic multi-component disk plus halo model, are more
complicated than sech2 or an exponential; and an analytical form for these has not been
obtained so far (see Section 4.2.6). Hence the results for the potential-density pairs from
[371] do not cover such a realistic system.

160
taken to be gravitationally coupled, and the disk is in the gravitational field
of the dark matter halo. The self-consistent vertical density distribution of
each disk component is obtained in terms of the input parameters: namely,
the surface density and velocity dispersion of the disk components; and the
halo mass distribution.
The core of the review consists of presenting this model and the results
from it. The inclusion of gas gravity and dark matter halo is shown to have
a crucial constraining effect on the the vertical structure of the stellar disk.
This model highlights the point that the vertical structure of stellar disk
cannot be treated in isolation as has been typically done in the past; rather,
the gravitational coupling with gas and the dark matter halo must be taken
into account to get the correct physical distribution for each disk component.
The review is written in a pedagogic style, stressing the underlying physics
and the mathematics in the formulation of the equations, and the treatment
to solve these. The development of equations for treating the different physi-
cal cases: namely, a thin disk case; the thick/low density disk case (when the
R − z coupling is taken into account); and the more general cases including
kinematic effects as done by using the complete Jeans equations, and a non-
isothermal velocity dispersion; has been described clearly, and the difference
between these cases is explained. The aim is to consider progressively more
detailed physics in the problem, and treat realistic cases; starting from the
simple, standard isothermal, one-component case that gives a sech2 density
profile. We hope that in view of the above features, this review will be useful
for a reader, to get an easy introduction to this important and active subject.

9.1. Main results from the multi-component disk plus halo model
1. The most significant result from the model is that the self-consistent
vertical density distribution of each disk component is constrained to be
closer towards the mid-plane: such that, the mid-plane density is higher; the
disk thickness or the HWHM of the distribution is smaller; and simultane-
ously the distribution is steeper compared to the one-component, isothermal
case. This is due to the inclusion of the additional gravitational force of the
other disk components and the dark matter halo in the coupled case (Sections
4.2.5, 4.2.6).
2. Due to the lower dispersion of gas, it has a smaller scale height than
the stars. Hence the gas contributes significantly to the force near the mid-
plane. Therefore, gas has a strong constraining effect on the stars despite its
small mass fraction in the disk. The stellar disk is constrained vertically in

161
the inner galactic disk mainly by the interstellar gas; and by the dark matter
halo in the outer disk. Thus, it is necessary to include both gas and the dark
matter halo to obtain an accurate vertical distribution of the stellar disk.
We stress that in both these regimes, the stellar distribution is steeper
than the sech2 profile obtained for a one-component, isothermal case. If the
stellar density profile is represented as sech2/n , then n > 1 in the inner region
where the gas has a dominant influence; while n < 1 in the outer region
where the halo is dominant. We point out that the latter case has not been
noted earlier in the literature (see Section 4.2.6). However, the value of n
obtained this way is not robust, hence the form sech2/n is not valid for a
realistic disk (Section 4.2.6).
3. The model simultaneously gives the self-consistent vertical density
distribution in each disk component. The results from this model explain
the observed scale heights in stars, HI and H2 in the Milky Way very well.
We stress that although the focus of this review is to study the stellar vertical
distribution; the results for the HI and H2 gas distributions are obtained at
the same time, and are an important feature of this model. The results also
naturally explain the observed trends in the density distribution of the thin
disk such as the steeper-than-sech2 profiles, and broad wings at high z (and
high R), as seen in external galaxies; without invoking a spurious second disk
as is done in the literature to explain the wings.
4. The self-gravitational energy of a multi-component disk is derived
using analytical techniques. Surprisingly, the potential energy per unit area
for a particular disk component remains the same in the coupled case (Section
4.6). However, the constrained distribution due to the higher gravitational
force at each z in the coupled case is predicted to make the disk robust
against distortion; say, by a tidal encounter with a satellite galaxy.
5. In the more general versions of the model, additional kinematical fea-
tures such as the tilt of the velocity ellipsoid and the cross terms in the Jeans
equations are included through the complete Jeans equations (Section 4.3).
Also, the observed non-isothermal velocity dispersion is taken into account
(Section 4.4). These are shown to shape the density profile which could
be substantially different compared to a sech2 profile that is obtained for a
single-component, isothermal case; with the mid-plane density and the disk
thickness (HWHM) that differ by ∼ 30 − 40%. The non-isothermal velocity
dispersion causes the density distribution to be more vertically extended and
broader at high z. It may be feasible to compare these predictions with data
from future Gaia data releases and other data, as well as with high-resolution

162
numerical simulations.
6. The disk vertical structure is routinely studied assuming an isothermal
velocity dispersion, for simplicity. While testing the limits of validity of this
assumption, it was shown [21] that even a small vertical velocity gradient of
∼ 2 − 3 km s−1 kpc−1 results in the mid-plane density being smaller by a
sizeable amount ∼ 10−14% compared to the isothermal, one-component case.
It is possible that such small gradients could arise locally in a galactic disk;
indeed, a strictly constant velocity dispersion denotes an idealized case. This
result is a warning that the sech2 law that is routinely used for its convenient,
analytical form may not be strictly correct even for a single-component disk.
7. A generic and robust result is that the typical stellar galactic disk,
which is in hydrostatic balance; and when using the observed input param-
eters, is shown to flare moderately, with the vertical scale height increasing
by a factor of 2-3 within the optical disk. This is true even for an isolated
galaxy, so an encounter is not necessary for a disk to flare. We stress that the
flaring is not due to the multi-component treatment; rather, it is seen in the
stars-alone disk as well (e.g., in the Milky Way - see Section 5.3 for details).
In fact, the presence of dark matter halo reduces the flaring to a moderate
value at large radii, as shown in the Milky Way [19]. A flaring disk is con-
trary to the long-held assumption of constant scale height in the literature
[22]; however, it is in agreement with and naturally explains the trend seen
in recent observations and simulations (see Section 5). Thus, future studies
of galactic structure must take account of the flaring of the stellar disk.
Necessity of using the multi-component disk model in future work:
In view of the above results, it is clear that the realistic, multi-component
disk plus halo model should be used for an accurate treatment in future stud-
ies of disk vertical structure. We stress that such a rigorous, self-consistent
calculation is not just of academic interest; rather, it is essential to use this
approach to obtain a physically correct vertical density distribution. The
model results clearly show that the resulting self-consistent vertical density
distribution for stars differs from sech2 and is steeper, in all realistic cases
considered; and a flaring stellar disk is shown to be a generic result.
In the literature, however, a simple sech2 distribution or its limiting form
at high z, namely, an exponential function (Section 3.1); along with a con-
stant scale height for stars, are routinely used for convenience (for example,

163
see the references in Section 6.4).21 The actual model density profile for a re-
alistic disk is more complex than given by either of these profiles (see Section
4.2.6).
We caution that the use of a sech2 or an exponential function does not
describe the true vertical density distribution for a real disk; and such us-
age may lead to dynamically incorrect results, or may make one miss the
underlying physics (e.g., Section 4.5.4).
On the other hand, applying the rigorous multi-component disk plus halo
model to theoretical problems or to analyze the observed data; though it
involves more work, can lead to new physical insights into the disk structure
(e.g., [160]; also, see Section 4.5.1). The above model [16] used to analyze the
S4 G data by [160] led to the interesting result about the relative masses in
the thin and thick disks being comparable, hence the conclusion of an early
origin of the thick disk. This new physical result would have been missed if
the standard practice of using sech2 for fitting the data had been employed,
as noted by the authors.
Overall, the mass distribution is vertically constrained in a coupled disk
– which will be reflected in the model density profile; and the effect of this
constraining on dynamical studies can only be taken into account by using
the self-consistent model density profile.
Note that a simple analytical form to represent the numerical density pro-
file, ρ(z), or the corresponding potential, resulting from the multi-component
disk plus halo model for any of the realistic cases studied, is not easy to pre-
scribe; and this has not been attempted so far (Sec 4.2.6, Appendix A),
although it would be convenient to have this.

9.2. Dynamical implications of the model

As seen above, the multi-component disk plus halo model significantly
changes the self-consistent vertical density distribution, and the thickness
of each disk component. The associated gravitational potential and the net
vertical force are also different compared to the one-component case. Thus,
apart from being interesting in themselves, the results for the disk vertical
structure have a direct bearing on the various dynamical processes in the
disk.

21
Also, note that the modern text-book titled ”Galaxies and Astrophysics of Galaxies”,
Bovy, J., Princeton University Press (a book under preparation) in fact recommends the
usage of an exponential stellar profile and constant scale height.

164
 Various examples of the dynamical applications have already been dis-
cussed in the previous sections (in particular, see Sections 4.5, 5 and 6 for
details): namely, star formation; disk stability; the feasibility and evolution
of spiral arms; and, the constraining of the density profile and shape of the
dark matter halo. Thus the results from this model have broader implications
for various fundamental properties of galaxies.

9.3. Unique features of study of disk vertical structure

The study of vertical structure of the disk has some unique features. In
keeping with the pedagogic nature of this review, it is worth summarising
these features, which were discussed in this review.
First, the self-consistent vertical density distribution in a disk depends
only on the local conditions at a given R. Thus the study of disk vertical
structure reduces to a local, one-dimensional problem along z. This is true
for a thin disk: for a one-component case (Section 3.1), and for a multi-
component disk (Section 4.2.1); as well as for a physically thick disk if it is
in a rotational equilibrium (Section 4.3.1). Thus, interestingly, the vertical
structure allows one to trace the local gravitational potential, unlike in the
planar case (see Section 3).
On the other hand, the entire vertical density distribution along the z
direction at a given radius has to be determined self-consistently by jointly
solving the Poisson equation and the equation of hydrostatic equilibrium.
Thus the determination of the vertical structure is a harder problem than
that of the galactic mass model where a test particle approach can be used
to determine the planar mass distribution from the observed rotation curve
(see Section 3 for details).
Second, due to the open nature of disk geometry, each disk component
responds to the potential of the other disk components, even those lying at
larger z values (see Appendix A in [121]). For example, the gas – despite
having a smaller scale height than stars – can respond to the stellar potential
(Section 4.2.1). This makes it possible to formulate a multi-component disk
plus halo model.
Third, since the disk vertical structure can be phrased as a local, one-
dimensional problem along z; the self-consistent density distribution in a
disk component in the coupled case is only affected by the local properties
of the other disk components and the halo, at a given R. Thus, the stellar
density distribution is affected by H2 gas in the inner disk where most of the

165
molecular gas is observed, and by the HI gas in the middle and outer Galaxy.
This trend will be true in all galaxies.
Fourth, the potential energy per unit area for a disk component is un-
changed even in a coupled case. The interaction acts as an internal force,
merely redistributing the energy along z. However, it takes more energy to
raise a unit mass by a certain distance along z in the coupled case, or the
gravitational potential is higher. Hence, the disk component is more tightly
bound and harder to distort than the one-component case (see Section 4.6).
Finally, the disk vertical structure contains clues regarding the formation
and evolution of the disk (Section 1).

9.4. Future problems and trends

There are a number of important, open problems in this field, covering
the range from direct model applications to dynamical implications of the
model. This topic was discussed in some detail, giving the motivation and
the detailed outline of each problem, in Section 8. A few of these are briefly
summarized here, starting with the direct model applications:
1. First, crucial input parameters such as stellar velocity dispersion values
in external galaxies can now be extracted from the IFU data, available from
various surveys. This can yield more accurate model results for the vertical
density profiles. This will also allow one to check and confirm whether the
stellar disk flares in general, see Section 5. Alternatively, the model results
could be taken as a template to be compared with new data such as Gaia
DR3, or other data that become available from future observations.
2. Second, the model could be applied to understand the vertical struc-
ture of the kinematically and chemically distinct Thick disk component. To
do this, the multi-component disk plus halo model has to applied in the
physically thick disk approach (see Sections 7.1, and 8 for details).
3. The data from JWST has revealed the surprising result that disk
galaxies were common, and even dominant, at a redshift as high as 4-6 [372,
373, 374]. These even showed spiral features, which indicates an underlying
rotating disk at a redshift of 4 ([375]; also, see [376]). This is an emergent
area of research now.
Thus, the time is now ripe for a systematic study of the vertical structure
of disks in high-redshift galaxies. Physically motivated density profiles for
a thin and a Thick disk can be obtained by applying the multi-component
disk plus halo model to such a system. Then, by comparing these density
profiles with the data from JWST, the best-fit parameters for the thin and

166
Thick disk can be extracted (see Section 8 for details). Additionally, a study
of disk vertical structure could shed light on the dynamics and evolution of
high redshift galaxies, which is not well-understood.
In addition, the overall constrained vertical distribution in a coupled case
can have various dynamical implications, which are worth pursuing:
1. First, it would be interesting to check by simulations, if the vertical
constraining that is obtained in a realistic, multi-component disk makes the
disk more robust against tidal distortion - as was predicted by [90]. This
would be particularly important for the early evolution of high-redshift, gas-
rich galaxies.
2. Second, the steeper-than-sech2 profile and a vertically constrained
distribution, as resulting from a multi-component disk plus halo model, is
likely to affect the properties of warps (see e.g., [258]) and other bending
modes. This topic should be investigated in detail.
3. Third, the 3-D HI data cube now available for galaxies can yield more
accurate observed scale height values and rotation curves. Hence, the new
approach developed (Section 6), where the above two are used as indepen-
dent constraints on the model, can now be applied systematically to a large
number of galaxies; so as to better constrain the shape and density profile of
their dark matter halos.

9.5. Summary and outlook

In the end, we would like to stress that the study of disk vertical struc-
ture is an ongoing progress: with many outstanding unsolved problems,
mainly triggered by new observations; and a growing realization that a multi-
component disk plus halo model is necessary to treat a realistic galactic disk.
This model, with the rigorous and self-consistent approach embodied in it,
would provide the necessary theoretical framework to physically understand
and interpret the new observations from Gaia, JWST, and other upcoming
future observatories. These include data on high redshift galaxies, which
have surprising and unexpected properties, as discussed above. This is a
new, relatively unexplored field, and the multi-component disk plus halo
model discussed in this review is a timely tool to analyze this data. This will
help us better understand the vertical structure of galactic disks, and thus
also help shed light on the dynamics and early evolution of galaxies.
The past experience from Gaia DR2 and RAVE shows that these led to
the discovery of unexpected features such as non-isothermal velocity disper-
sion (Section 4.4.2), and the enigmatic phase-space spirals (Section 7.2.3).

167
The latter indicate that a fraction of the disk could be out-of-equilibrium,
which is an active topic of research now. We can similarly expect that data
from JWST and other future observations are likely to reveal more surprises,
that will open up new theoretical topics in the field. The study of disk verti-
cal structure is thus an exciting topic of study now, and promises to remain
so in the coming years.
We hope that this review will provide an in-depth introduction to the
important topic of the vertical structure and dynamics of a galactic disk;
along with an outline of some of the open, challenging problems in it. This
would, hopefully, enthuse more people to work in this field and to explore its
rich physics further.

Acknowledgments:
During the past two decades, I have had the great pleasure of working with
many students, and other collaborators, on the topic of the vertical structure
of a galactic disk and related dynamics, and in particular, on developing the
multi-component disk plus halo model and working out its various applica-
tions. It is a pleasure to thank my students; in chronological order of our
work together, they are: Chaitra Narayan, Kanak Saha, Pratyush Pranav,
Arunima Banerjee, Rathulnath R., Soumavo Ghosh, and Suchira Sarkar.
Special thanks are due to Chaitra and Suchira for extensive collaboration;
also to Chaitra for being there at the beginning of this study, and to Suchira
for several useful discussions during the writing of this review. I also thank
the various collaborators; in chronological order, they are: Leo Blitz, Evan
Levine, Lynn Matthews, Elias Brinks, Ioannis Bagetakos, Nirupam Roy, and
Meera Nandakumar. Finally, I would like to thank friends who are physicists
but not astronomers: namely, (the late) Rohini Godbole for appreciating the
results from this model over the years; and Suman Iyer for giving a patient
hearing during the writing of this review.
I am happy to acknowledge the support from INSA, New Delhi as an INSA
Senior Scientist for the past two years (2023-2025), and also from DST, New
Delhi for a J.C. Bose National Fellowship for an extended period in the past
during a large part of the work reported here.

168
Acronyms
CALIFA Survey: Calar Alto Legacy Integral Field Area Survey
ISM: Interstellar Medium
IFU: Integral Field Unit
JWST: James Webb Space Telescope
LAMOST: The Large Sky Area Multi-Object Fiber Spectroscopic Telescope
SDSS: Sloane Digital Sky Survey
S4 G: Spitzer Survey of Spiral Structure in Galaxies

169
 Table 1: Glossary of important terms used:
Symbol used Description of the physical parameter

R, ϕ, z Galactic cylindrical coordinates, where R is the

Galactocentric radius and z is normal to the Galactic plane
ρ(z)i Mass density as a function of z, with the subscript
i=s,HI,H2 ,h corresponding to stars, interstellar HI, and H2
gas, and dark matter halo, respectively
Σi (R) Surface density at a given R for a disk component, with
the subscript i=s,HI,H2 , as defined above.
Φi Gravitational potential, with the subscript i=s,HI,H2 ,h as
defined above.
⟨(vz2 )i ⟩1/2 = (σz )i R.M.S. velocity dispersion along z, with the
subscript i=s,HI,H2 as defined above.
(Kz )i Force per unit mass along the z direction, with the subscript
i=s,HI,H2 ,h as defined above.
RD , Rv The scale length for exponential fall-off with radius of stellar
disk surface density and vertical velocity dispersion, respectively
ρ 0 , z0 The mid-plane disk density and the parameter indicating
the scale height in an isothermal, one-component, thin disk
distribution given by sech2 (z)
ρ0d The mid-plane disk density in a general, coupled system
HWHM The half-width at half maximum (HWHM) of the vertical
density distribution, also referred to as the scale height
of the vertical density distribution
ρ0h , Rc The mass density of the dark matter halo at the centre of a
galaxy, and the core radius of the halo, respectively for a
psuedo-isothermal density distribution
q = c/a Shape of the spheroid, where c and a correspond to the
the vertical and the radial major axis of the spheroid,
respectively. q > 1, < 1 corresponds to a prolate and an
oblate spheroid respectively. q = 1 represents a sphere with
a = b = c where b is the semi-minor axis in the disk plane.
W Gravitational potential energy per unit area for a disk

170
10. Appendix A. Thickness of a self-gravitating disk and disk ver-
tical density profile
A.1 Thickness of a self-gravitating disk
For a disk mass distribution, its vertical density profile, ρ(z), and thick-
ness are important physical properties that characterize the disk and also
provide a handy way of comparing various disks. However, for a gravitating
system, neither its radial extent nor its thickness is sharp or well-defined.
Though this is an important point, this is not often mentioned clearly in the
literature. Further, a different definition of thickness is employed for different
cases. The aim of this Appendix is to point this out and bring some clarity
and uniformity to the definition of disk thickness that has been employed in
the literature for different density profiles.
The radial extent of disk of an external galaxy, is taken to be given by the
radius of the optical disk or the visible extent of the stellar disk. The radius
of the optical disk is typically defined to be the radius of an isophote of a
given limiting magnitude, such as 26.5 mag arc sec−2 (known as the Holmberg
radius, RH ); or 25 mag arc sec−2 , corrected for inclination (known as the de
Vaucouleurs radius), see [2]. For the typical values of central intensity of a
spiral galaxy, and the sky brightness, RH ∼ 4 − 5RD (e.g., [205]).22
It is clear that the functional form of the vertical density profile, or specif-
ically the rate of fall-off of the density distribution, would be important in
defining its vertical extent as well as what can be taken to be its thickness.
So the density profile and the thickness are directly coupled.
For example, the observed stellar distribution (based on the star counts)
close to the Galactic mid-plane is typically fitted by an exponential distri-
bution and hence the rate of exponential fall-off of density,zexp , is taken to
denote the disk thickness or scale height; and its value in ∼ 300 pc in the
solar neighbourhood (e.g., [86]). Another way to estimate the characteristic
thickness is to divide the total surface density by the mid-plane mass density
of stars; and this gives a value of 350 pc in the solar neighbourhood [2]. The

22
The above definition for the radius of the optical disk depends on the sky brightness
as well as the technological capabilities of detecting faint regions [377]. The actual radial
extent of the mass distribution in a galactic disk could be much larger. This is confirmed
by the detection of faint, highly extended outer regions of some galaxies using modern
sensitive telescopes: for example, the detection by the Dragonfly Telescope Array of outer,
faint disks up to ∼ 18RD in M101 [378] and up to ∼ 23RD in NGC 2841 [379].

171
interstellar HI gas typically is taken to obey a Gaussian distribution, and the
half-width-at-half-maximum (HWHM) of this distribution is taken to define
the gas thickness [108, 67]. For HI gas, this is measured to be ∼ 150pc in the
solar neighbourhood [42, 67].
Theoretically, a gravitating, one-component, isothermal, thin disk distri-
bution has a form sech2 (z/z0 ) along z ([12]; also, see Section 3.1). Here z0
is defined to be the scale parameter, as per the notation used by [12, 13].
Later, in the literature, the quantity z0 has often been refereed to as the
”scale height” [8], and is taken to be an indicator of disk thickness.
A.2 HWHM of vertical density distribution as the thickness
In general, the density distribution of stars in a galactic disk deviates from
the sech2 function predicted for an isothermal, one-component disk. This is
seen observationally for external galaxies (e.g., [14]); and, also found theo-
retically for the self-consistent density distribution resulting from a multi-
component disk plus halo model (see Sections 4.1 to 4.4, in particular, see
[19]). For each disk component in the coupled case, the vertical density distri-
bution is more concentrated towards the mid-plane, and is steeper, than the
sech2 distribution obtained for the corresponding one-component case. Thus,
for a constant surface density (i.e., at a given R), the additional gravitational
force due to gas and the halo results in a steeper profile, and simultaneously
a smaller disk thickness for stars (see Section 4.2.5). The non-isothermal
velocity dispersions tends to have the opposite effect of making the density
profile broader than sech2 , especially at high z- the latter appears as a wing
(Section 4.4.2).
Thus, even though each isothermal disk component by itself would satisfy
a sech2 distribution; the resulting stellar vertical density distribution in the
coupled system is completely different from sech2 , see Section 4.2.6 for details.
In these realistic cases, it is not meaningful to use z0 as the indicator of
disk thickness. Hence, instead, the HWHM (Half Width at Half Maximum)
of the vertical density distribution of a disk component was introduced as
an indicator of the scale height disk thickness: by [112] while treating the
effect of a cloud complex on the disk; and by [16] while treating a multi-
component disk plus halo model. The use of HWHM was in analogy to what
is routinely done in the study of ISM [67]. The HWHM was also refereed to
as the scale height (or disk thickness) in the above two papers and in other
subsequent papers on this topic. It should be pointed out that the definition
of the HWHM takes account of the density variation with z; but it does not

172
explicitly depend on the actual functional form of the density distribution.
Hence HWHM is a robust indicator of disk thickness. In addition, it is easy
to determine its value from the model results as well as from the observed
data.
The subsequent detailed work on the multi-component disk plus halo
model showed that in all realistic cases the resulting density profile, ρ(z),
differs from sech2 ; and could instead be attempted to be specified by a more
general functional form, sech2/n ([14]; or, see Eq. 4.7 ). However, it is found
that here the exponent n is not robust; rather, it varies with ∆z, the range
of z over which the density distribution is fitted by the above functional form
(see Section 4.2.6). Thus n as defined by [14] is not a useful indicator of the
steepness of the vertical density profile; and Eq. (4.7) as proposed by [14] is
not valid for a realistic galactic disk.
This behaviour was shown for the various physical cases considered: namely,
the multi-component disk plus halo model, for a thin disk as applied to the
Milky way [19]; and for a thick disk as applied to UGC 7321 [91]; and also
for one-component thick disk, with kinematic effects included (Model B re-
spectively from [20]); and the non-isothermal case [21]. Therefore, defining a
scale height based on the functional form of the density distribution (analo-
gous to z0 for the sech2 distribution) would not be feasible. Hence, the use of
HWHM introduced and used by [112, 16] to denote the scale height or disk
thickness makes good practical sense. This usage was then also adopted in
the above-mentioned subsequent papers. The HWHM from the above physi-
cally obtained density distributions is a robust indicator of disk thickness or
scale height.
Therefore, we warn that in most realistic cases, simply assuming a sech2
vertical density distribution, or taking the high z limit of this – namely,
an exponential distribution – and taking the associated z0 parameter (see
Eq.(3.5)) as the scale height, and further assuming it to be constant, as is
routinely done in the literature, does not represent the true, detailed physics
in the problem. Such usage would lead to dynamically incorrect results.
Finally, we comment on a point made by ([170]); namely, that the param-
eter n ̸= 1 in [14] (n as defined by Eq. (4.7) corresponds to a non-isothermal
case. We point out that this is not strictly true. As shown above, any
deviation from a one-component, isothermal case (including the isothermal,
multi-component case as well as a non-isothermal, one-component case) gives
rise to a density profile that has n ̸= 1. In any case, as described above, the
parameter n obtained for a general case is not robust. This is because Eq.

173
(4.7) proposed by [14] is not adequate to explain the results obtained for a
realistic disk (see the discussion in Section 4.2.6).
As discussed above, most realistic cases give a density profile different
than sech2 . We point out that a simple analytical form for the actual density
profile resulting from the multi-component disk plus halo model for any of the
realistic cases studied is not easy to prescribe, and has not been attempted
so far (Sec 4.2.6); although such an expression would be useful.
Vertical profile of gas density distribution
A distribution of any self-gravitating disk component will be more con-
strained closer to the mid-plane in the coupled case, and would show a dis-
tribution steeper than sech2 (Section 4.2.5). The HWHM of the density
distribution is used to denote the thickness for each of the disk components,
namely, stars, HI and H2 [16], also see the above discussion.
[131] states that the gas distribution in the coupled case is a Gaussian
(see Fig. 7 in [131]). However, that paper has not done a mathematical
fit to check this or to see whether the fit to n (in sech2/n ) is a function of
z. In any case, a fit close to a Gaussian for gas is not surprising. It is
well-known (e.g., [87]) that in the potential of the main mass component of
the disk, namely the stars, the gas (when treated as massless particles) has
a Gaussian distribution (as given by Eq.(3.6)). In the study of ISM, the
HI distribution is routinely fitted by a Gaussian ([67];also, see Section 3.2).
However, in the more realistic, coupled case, since it includes gas gravity, the
gas distribution would be expected to be steeper than this. Certainly the HI
gas distribution in the field of dark matter halo and stars is much steeper
than the gas-alone case - the latter under its own self-gravity (as seen in
Section 4.2.5, Fig. 15). A quantitative measurement of the steepness of gas
vertical density distribution in the coupled case needs to be done, to ensure
a full understanding.
A.3 Half-mass scale height, z1/2 , as an indicator of disk thickness:
Another indicator of disk thickness was introduced by [19], namely, z1/2 ,
which they called the half-mass scale height, measured from the mid-plane.
This is defined to contain half the mass in a column of unit area at a given
radius, or half the total stellar surface density at a given radius. This will
also implicitly depend on the shape of the density profile at larger z values,
unlike the value of HWHM. Hence a smaller z1/2 indicates a distribution that
is concentrated closer to the mid-plane. This parameter has not been noted or
used much in the literature; however, see [129] who independently defined a

174
similar quantity to define thickness during the analysis of TNG50 simulations
data. From an observational viewpoint, z1/2 can be thought of as an indicator
of the half-light scale-height of an edge-on galaxy with M/L considered to be
constant. That is, z1/2 is the height that contains half the light from a disk
in a column density at a given radius. The two definitions (HWHM and z1/2 )
give values of heights that are comparable, with the HWHM being somewhat
larger than the z1/2 values (see Tables 2 and 3 from [19]).
A.4 Some special cases of density profiles:
In some special cases, a simple analytical function is a reasonably good
fit to the vertical density distribution, and is used for simplicity. The conver-
sion between the HWHM and the scale height corresponding to a particular
functional form, for some typical cases, is given next. For example, consider
a sech2 distribution (proportional to sech2 (z/z0 ))); a sech distribution (pro-
portional to sech(z/zsech )), and an exponential distribution (proportional to
exp(−z/zexp )). In these cases, the relation between the corresponding scale-
factors and the HWHM is given respectively as: z0 = HMHM/0.88; zsech =
HWHM/0.759 ; and zexp = HWHM/0.693.

175
References
[1] A. Sandage, The Hubble Atlas of Galaxies, 1961.

[2] J. Binney, S. Tremaine, Galactic dynamics, 1987.

[3] E. P. Hubble, Realm of the Nebulae, 1936.

[4] Y. Sofue, V. Rubin, Rotation Curves of Spiral Galaxies, Annu. Rev.

Astron. Astrophys. 39 (2001) 137–174. doi:10.1146/annurev.astro.
39.1.137. arXiv:astro-ph/0010594.

[5] J. H. Oort, Stellar Dynamics, in: A. Blaauw, M. Schmidt (Eds.),

Galactic structure. Edited by Adriaan Blaauw and Maarten Schmidt.
Published by the University of Chicago Press, 1965, p. 455.

[6] H. Mo, F. C. van den Bosch, S. White, Galaxy Formation and Evolu-
tion, 2010.

[7] J. Silk, A. Di Cintio, I. Dvorkin, Galaxy Formation, in: Proceedings

of the International School of Physics ’Enrico Fermi’ Course 186 ’New
Horizons for Observational Cosmology’ Vol. 186, volume 186, 2014, pp.
137–187. doi:10.3254/978-1-61499-476-3-137. arXiv:1312.0107.

[8] P. C. van der Kruit, K. C. Freeman, Galaxy Disks,

Annu. Rev. Astron. Astrophys. 49 (2011) 301–371. doi:10.1146/
annurev-astro-083109-153241. arXiv:1101.1771.

[9] J. Bland-Hawthorn, O. Gerhard, The Galaxy in Context: Structural,

Kinematic, and Integrated Properties, Annu. Rev. Astron. Astrophys.
54 (2016) 529–596. doi:10.1146/annurev-astro-081915-023441.
arXiv:1602.07702.

[10] A. J. Deason, V. Belokurov, Galactic Archaeology with Gaia, New

Astronomy Review 99 (2024) 101706. doi:10.1016/j.newar.2024.
101706. arXiv:2402.12443.

[11] J. A. S. Hunt, E. Vasiliev, Milky Way dynamics in light of Gaia, arXiv

e-prints (2025) arXiv:2501.04075. doi:10.48550/arXiv.2501.04075.
arXiv:2501.04075.

176
[12] J. Spitzer, Lyman, The Dynamics of the Interstellar Medium. III.
Galactic Distribution., Astrophys. J. 95 (1942) 329. doi:10.1086/
144407.

[13] G. L. Camm, Self-gravitating star systems, Mon. Not. R. Astron. Soc.

110 (1950) 305. doi:10.1093/mnras/110.4.305.

[14] P. C. van der Kruit, The three-dimensional distribution of light and

mass in disks of spiral galaxies., Astron. Astrophys. 192 (1988) 117–
127.

[15] A. Barteldrees, R. J. Dettmar, Parameters of stellar disks from CCD

surface photometry of edge-on galaxies, Astron. Astrophys. Supple-
ment 103 (1994) 475–502.

[16] C. A. Narayan, C. J. Jog, Vertical scaleheights in a grav-

itationally coupled, three-component Galactic disk, Astron.
Astrophys. 394 (2002) 89–96. doi:10.1051/0004-6361:20021128.
arXiv:astro-ph/0210239.

[17] P. M. W. Kalberla, Dark Matter in the Milky Way. I. The Isothermal

Disk Approximation, Astrophys. J. 588 (2003) 805–823. doi:10.1086/
374330.

[18] A. Banerjee, C. J. Jog, The Origin of Steep Vertical Stellar Distribution

in the Galactic Disk, Astrophys. J. 662 (2007) 335–340. doi:10.1086/
517605. arXiv:astro-ph/0703169.

[19] S. Sarkar, C. J. Jog, The constraining effect of gas and the

dark matter halo on the vertical stellar distribution of the Milky
Way, Astron. Astrophys. 617 (2018) A142. doi:10.1051/0004-6361/
201833510. arXiv:1808.02047.

[20] S. Sarkar, C. J. Jog, General model of vertical distribution of

stars in the Milky Way using complete Jeans equations, Mon. Not.
R. Astron. Soc. 492 (2020) 628–633. doi:10.1093/mnras/stz3512.
arXiv:1912.08829.

[21] S. Sarkar, C. J. Jog, Vertical stellar density distribution in a non-

isothermal galactic disc, Mon. Not. R. Astron. Soc. 499 (2020) 2523–
2533. doi:10.1093/mnras/staa2924. arXiv:2009.10097.

177
[22] P. C. van der Kruit, L. Searle, Surface photometry of edge-on spiral
galaxies. I - A model for the three-dimensional distribution of light in
galactic disks., Astron. Astrophys. 95 (1981) 105–115.
[23] Y. Jia, C. Yang, Y. Chen, C. Du, G. Zhao, Investigating the ver-
tical distribution of the disk as a function of radial action: Results
from simulations, Astron. Astrophys. 692 (2024) A167. doi:10.1051/
0004-6361/202452294. arXiv:2411.12432.
[24] L. Thulasidharan, E. D’Onghia, R. Benjamin, R. Drimmel, E. Pog-
gio, A. Queiroz, The Age-Thickness Relation of the Milky
Way Disk: A Tracer of Galactic Merging History, arXiv e-
prints (2024) arXiv:2412.12304. doi:10.48550/arXiv.2412.12304.
arXiv:2412.12304.
[25] J. H. Oort, Note on the determination of Kz and on the mass density
near the Sun., B.A.N. 15 (1960) 45.
[26] D. Burstein, Structure and origin of S0 galaxies. III. The luminosity
distribution perpendicular to the plane of the disks in S0’s., Astrophys.
J. 234 (1979) 829–836. doi:10.1086/157563.
[27] K. Sheth, M. Regan, J. L. Hinz, A. Gil de Paz, K. Menéndez-Delmestre,
J.-C. Muñoz-Mateos, M. Seibert, T. Kim, E. Laurikainen, H. Salo,
D. A. Gadotti, J. Laine, T. Mizusawa, L. Armus, E. Athanassoula,
A. Bosma, R. J. Buta, P. Capak, T. H. Jarrett, D. M. Elmegreen,
B. G. Elmegreen, J. H. Knapen, J. Koda, G. Helou, L. C. Ho, B. F.
Madore, K. L. Masters, B. Mobasher, P. Ogle, C. Y. Peng, E. Schin-
nerer, J. A. Surace, D. Zaritsky, S. Comerón, B. de Swardt, S. E. Meidt,
M. Kasliwal, M. Aravena, The Spitzer Survey of Stellar Structure in
Galaxies (S4G), Publications of the Astronomical Society of the Pacific
122 (2010) 1397. doi:10.1086/657638. arXiv:1010.1592.
[28] P. C. van der Kruit, L. Searle, Surface photometry of edge-on spi-
ral galaxies. II - The distribution of light and colour in the disk and
spheroid of NGC 891., Astron. Astrophys. 95 (1981) 116–126.
[29] R. de Grijs, R. F. Peletier, The shape of galaxy disks: how the
scale height increases with galactocentric distance., Astron. Astro-
phys. 320 (1997) L21–L24. doi:10.48550/arXiv.astro-ph/9702215.
arXiv:astro-ph/9702215.

178
[30] K. C. Freeman, On the Disks of Spiral and S0 Galaxies, Astrophys. J.
160 (1970) 811. doi:10.1086/150474.

[31] L. Spitzer, Physical processes in the interstellar medium, 1978. doi:10.

1002/9783527617722.

[32] B. T. Draine, Physics of the Interstellar and Intergalactic Medium,

2011.

[33] M. Kregel, P. C. van der Kruit, R. de Grijs, Flattening and trunca-

tion of stellar discs in edge-on spiral galaxies, Mon. Not. R. Astron.
Soc. 334 (2002) 646–668. doi:10.1046/j.1365-8711.2002.05556.x.
arXiv:astro-ph/0204154.

[34] S. P. C. Peters, P. C. van der Kruit, R. J. Allen, K. C. Freeman, The

shape of dark matter haloes - V. Analysis of observations of edge-on
galaxies, Mon. Not. R. Astron. Soc. 464 (2017) 65–84. doi:10.1093/
mnras/stw2101. arXiv:1605.08209.

[35] C. A. Narayan, C. J. Jog, Origin of radially increasing stellar scale-

height in a galactic disk, Astron. Astrophys. Lett. 390 (2002) L35–L38.
doi:10.1051/0004-6361:20020961. arXiv:astro-ph/0207453.

[36] K. Rohlfs, H. J. Wiemer, On the disk-thickness of spiral galaxies,

Astron. Astrophys. 112 (1982) 116–119.

[37] A. V. Kasparova, I. Y. Katkov, I. V. Chilingarian, An excessively

massive thick disc of the enormous edge-on lenticular galaxy NGC 7572,
Mon. Not. R. Astron. Soc. 493 (2020) 5464–5478. doi:10.1093/mnras/
staa611. arXiv:1912.04887.

[38] E. M. Xilouris, Y. I. Byun, N. D. Kylafis, E. V. Paleologou, J. Papa-

mastorakis, Are spiral galaxies optically thin or thick?, Astron. Astro-
phys. 344 (1999) 868–878. doi:10.48550/arXiv.astro-ph/9901158.
arXiv:astro-ph/9901158.

[39] S. Vig, S. K. Ghosh, D. K. Ojha, Warp signatures of the Galactic disk

as seen in mid infrared from Midcourse Space Experiment, Astron.
Astrophys. 436 (2005) 867–877. doi:10.1051/0004-6361:20052920.
arXiv:astro-ph/0503388.

179
[40] M. López-Corredoira, A. Cabrera-Lavers, F. Garzón, P. L. Hammers-
ley, Old stellar Galactic disc in near-plane regions according to 2MASS:
Scales, cut-off, flare and warp, Astron. Astrophys. 394 (2002) 883–899.
doi:10.1051/0004-6361:20021175. arXiv:astro-ph/0208236.

[41] M. López-Corredoira, J. Molgó, Flare in the Galactic stellar outer disc

detected in SDSS-SEGUE data, Astron. Astrophys. 567 (2014) A106.
doi:10.1051/0004-6361/201423706. arXiv:1405.7649.

[42] J. G. A. Wouterloot, J. Brand, W. B. Burton, K. K. Kwee, IRAS

sources beyond the solar circle. II. Distribution in the galactic warp.,
Astron. Astrophys. 230 (1990) 21–36.

[43] E. Florido, E. Battaner, A. Guijarro, F. Garzón, J. Jiménez-Vicente,

Near infrared observations of the truncation of stellar disks, As-
tron. Astrophys. 378 (2001) 82–96. doi:10.1051/0004-6361:20011190.
arXiv:astro-ph/0109295.

[44] E. Florido, E. Battaner, A. Guijarro, F. Garzón, A. Castillo-Morales,

Truncated stellar disks in the near infrared. I. Observations, Astron.
Astrophys. 455 (2006) 467–473. doi:10.1051/0004-6361:20053310.
arXiv:astro-ph/0609384.

[45] M. Pohlen, S. Zaroubi, R. F. Peletier, R.-J. Dettmar, On the three-

dimensional structure of edge-on disc galaxies, Mon. Not. R. Astron.
Soc. 378 (2007) 594–616. doi:10.1111/j.1365-2966.2007.11790.x.
arXiv:astro-ph/0703768.

[46] S. Bianchi, The dust distribution in edge-on galaxies. Radiative transfer

fits of V and K’-band images, Astron. Astrophys. 471 (2007) 765–773.
doi:10.1051/0004-6361:20077649. arXiv:0705.1471.

[47] G. De Geyter, M. Baes, P. Camps, J. Fritz, I. De Looze, T. M. Hughes,

S. Viaene, G. Gentile, The distribution of interstellar dust in CALIFA
edge-on galaxies via oligochromatic radiative transfer fitting, Mon.
Not. R. Astron. Soc. 441 (2014) 869–885. doi:10.1093/mnras/stu612.
arXiv:1403.7527.

[48] S. S. Savchenko, D. M. Poliakov, A. V. Mosenkov, A. A. Smirnov, A. A.

Marchuk, V. B. Il’in, G. A. Gontcharov, J. Seguine, M. Baes, The

180
 problem of dust attenuation in photometric decomposition of edge-on
galaxies and possible solutions, Mon. Not. R. Astron. Soc. 524 (2023)
4729–4745. doi:10.1093/mnras/stad2189. arXiv:2309.06257.

[49] K. A. Herrmann, R. Ciardullo, Planetary Nebulae in Face-On Spiral

Galaxies. III. Planetary Nebula Kinematics and Disk Mass, Astro-
phys. J. 705 (2009) 1686–1703. doi:10.1088/0004-637X/705/2/1686.
arXiv:0910.0266.

[50] Y. Momany, S. Zaggia, G. Gilmore, G. Piotto, G. Carraro, L. R. Bedin,

F. de Angeli, Outer structure of the Galactic warp and flare: explaining
the Canis Major over-density, Astron. Astrophys. 451 (2006) 515–538.
doi:10.1051/0004-6361:20054081. arXiv:astro-ph/0603385.

[51] D. Ferguson, S. Gardner, B. Yanny, Milky Way Tomography with

K and M Dwarf Stars: The Vertical Structure of the Galactic
Disk, Astrophys. J. 843 (2017) 141. doi:10.3847/1538-4357/aa77fd.
arXiv:1706.01900.

[52] H.-F. Wang, C. Liu, Y. Xu, J.-C. Wan, L. Deng, Mapping the Milky
Way with LAMOST- III. Complicated spatial structure in the outer
disc, Mon. Not. R. Astron. Soc. 478 (2018) 3367–3379. doi:10.1093/
mnras/sty1058. arXiv:1804.10485.

[53] J. Bovy, Stellar inventory of the solar neighbourhood using Gaia DR1,
Mon. Not. R. Astron. Soc. 470 (2017) 1360–1387. doi:10.1093/mnras/
stx1277. arXiv:1704.05063.

[54] P. S. Dobbie, S. J. Warren, A Bayesian Approach to the Vertical Struc-

ture of the Disk of the Milky Way, The Open Journal of Astrophysics
3 (2020) 5. doi:10.21105/astro.2003.05757. arXiv:2003.05757.

[55] Ó. Jiménez-Arranz, D. Horta, R. P. van der Marel, D. Nidever, C. F. P.

Laporte, E. Patel, H. W. Rix, Vertical structure and kinematics of the
LMC disc from SDSS/Gaia, arXiv e-prints (2025) arXiv:2501.04616.
doi:10.48550/arXiv.2501.04616. arXiv:2501.04616.

[56] J. R. Lewis, K. C. Freeman, Kinematics and Chemical Properties of

the Old Disk in the Galaxy, The Astronomical Journal 97 (1989) 139.
doi:10.1086/114963.

181
[57] R. Bottema, The stellar kinematics of galactic disks., Astron. Astro-
phys. 275 (1993) 16–36.

[58] M. A. Bershady, M. A. W. Verheijen, R. A. Swaters, D. R. Andersen,

K. B. Westfall, T. Martinsson, The DiskMass Survey. I. Overview, As-
trophys. J. 716 (2010) 198–233. doi:10.1088/0004-637X/716/1/198.
arXiv:1004.4816.

[59] S. F. Sánchez, R. C. Kennicutt, A. Gil de Paz, G. van de Ven,

J. M. Vı́lchez, L. Wisotzki, C. J. Walcher, D. Mast, J. A. L. Aguerri,
S. Albiol-Pérez, A. Alonso-Herrero, J. Alves, J. Bakos, T. Bartáková,
J. Bland-Hawthorn, A. Boselli, D. J. Bomans, A. Castillo-Morales,
C. Cortijo-Ferrero, A. de Lorenzo-Cáceres, A. Del Olmo, R. J. Dettmar,
A. Dı́az, S. Ellis, J. Falcón-Barroso, H. Flores, A. Gallazzi, B. Garcı́a-
Lorenzo, R. González Delgado, N. Gruel, T. Haines, C. Hao, B. Huse-
mann, J. Iglésias-Páramo, K. Jahnke, B. Johnson, B. Jungwiert,
V. Kalinova, C. Kehrig, D. Kupko, Á. R. López-Sánchez, M. Lyuben-
ova, R. A. Marino, E. Mármol-Queraltó, I. Márquez, J. Masegosa,
S. Meidt, J. Mendez-Abreu, A. Monreal-Ibero, C. Montijo, A. M.
Mourão, G. Palacios-Navarro, P. Papaderos, A. Pasquali, R. Peletier,
E. Pérez, I. Pérez, A. Quirrenbach, M. Relaño, F. F. Rosales-
Ortega, M. M. Roth, T. Ruiz-Lara, P. Sánchez-Blázquez, C. Sengupta,
R. Singh, V. Stanishev, S. C. Trager, A. Vazdekis, K. Viironen, V. Wild,
S. Zibetti, B. Ziegler, CALIFA, the Calar Alto Legacy Integral Field
Area survey. I. Survey presentation, Astron. Astrophys. 538 (2012) A8.
doi:10.1051/0004-6361/201117353. arXiv:1111.0962.

[60] J. Falcón-Barroso, M. Lyubenova, G. van de Ven, J. Mendez-Abreu,

J. A. L. Aguerri, B. Garcı́a-Lorenzo, S. Bekeraité, S. F. Sánchez,
B. Husemann, R. Garcı́a-Benito, D. Mast, C. J. Walcher, S. Zibetti,
J. K. Barrera-Ballesteros, L. Galbany, P. Sánchez-Blázquez, R. Singh,
R. C. E. van den Bosch, V. Wild, L. Zhu, J. Bland-Hawthorn, R. Cid
Fernandes, A. de Lorenzo-Cáceres, A. Gallazzi, R. M. González Del-
gado, R. A. Marino, I. Márquez, E. Pérez, I. Pérez, M. M. Roth,
F. F. Rosales-Ortega, T. Ruiz-Lara, L. Wisotzki, B. Ziegler, CAL-
IFA Collaboration, Stellar kinematics across the Hubble sequence in
the CALIFA survey: general properties and aperture corrections, As-
tron. Astrophys. 597 (2017) A48. doi:10.1051/0004-6361/201628625.
arXiv:1609.06446.

182
[61] K. B. Westfall, M. Cappellari, M. A. Bershady, K. Bundy, F. Belfiore,
X. Ji, D. R. Law, A. Schaefer, S. Shetty, C. A. Tremonti, R. Yan,
B. H. Andrews, J. R. Brownstein, B. Cherinka, L. Coccato, N. Drory,
C. Maraston, T. Parikh, J. R. Sánchez-Gallego, D. Thomas, A.-
M. Weijmans, J. Barrera-Ballesteros, C. Du, D. Goddard, N. Li,
K. Masters, H. J. Ibarra Medel, S. F. Sánchez, M. Yang, Z. Zheng,
S. Zhou, The Data Analysis Pipeline for the SDSS-IV MaNGA IFU
Galaxy Survey: Overview, The Astronomical Journal 158 (2019) 231.
doi:10.3847/1538-3881/ab44a2. arXiv:1901.00856.

[62] K. Bundy, M. A. Bershady, D. R. Law, R. Yan, N. Drory, N. MacDon-

ald, D. A. Wake, B. Cherinka, J. R. Sánchez-Gallego, A.-M. Weijmans,
D. Thomas, C. Tremonti, K. Masters, L. Coccato, A. M. Diamond-
Stanic, A. Aragón-Salamanca, V. Avila-Reese, C. Badenes, J. Falcón-
Barroso, F. Belfiore, D. Bizyaev, G. A. Blanc, J. Bland-Hawthorn,
M. R. Blanton, J. R. Brownstein, N. Byler, M. Cappellari, C. Con-
roy, A. A. Dutton, E. Emsellem, J. Etherington, P. M. Frinchaboy,
H. Fu, J. E. Gunn, P. Harding, E. J. Johnston, G. Kauffmann, K. Kine-
muchi, M. A. Klaene, J. H. Knapen, A. Leauthaud, C. Li, L. Lin,
R. Maiolino, V. Malanushenko, E. Malanushenko, S. Mao, C. Maraston,
R. M. McDermid, M. R. Merrifield, R. C. Nichol, D. Oravetz, K. Pan,
J. K. Parejko, S. F. Sanchez, D. Schlegel, A. Simmons, O. Steele,
M. Steinmetz, K. Thanjavur, B. A. Thompson, J. L. Tinker, R. C. E.
van den Bosch, K. B. Westfall, D. Wilkinson, S. Wright, T. Xiao,
K. Zhang, Overview of the SDSS-IV MaNGA Survey: Mapping nearby
Galaxies at Apache Point Observatory, Astrophys. J. 798 (2015) 7.
doi:10.1088/0004-637X/798/1/7. arXiv:1412.1482.

[63] S. M. Croom, M. S. Owers, N. Scott, H. Poetrodjojo, B. Groves, J. van

de Sande, T. M. Barone, L. Cortese, F. D’Eugenio, J. Bland-Hawthorn,
J. Bryant, S. Oh, S. Brough, J. Agostino, S. Casura, B. Catinella,
M. Colless, G. Cecil, R. L. Davies, M. J. Drinkwater, S. P. Driver,
I. Ferreras, C. Foster, A. Fraser-McKelvie, J. Lawrence, S. K. Leslie,
J. Liske, Á. R. López-Sánchez, N. P. F. Lorente, R. McElroy, A. M.
Medling, D. Obreschkow, S. N. Richards, R. Sharp, S. M. Sweet, D. S.
Taranu, E. N. Taylor, E. Tescari, A. D. Thomas, J. Tocknell, S. P.
Vaughan, The SAMI Galaxy Survey: the third and final data release,

183
 Mon. Not. R. Astron. Soc. 505 (2021) 991–1016. doi:10.1093/mnras/
stab229. arXiv:2101.12224.

[64] M. A. Bershady, K. B. Westfall, S. Shetty, D. R. Law, M. Cappellari,

N. Drory, K. Bundy, R. Yan, Asymmetric drift in MaNGA: mass
and radially dependent stratification rates in galaxy discs, Mon. Not.
R. Astron. Soc. 531 (2024) 1592–1612. doi:10.1093/mnras/stae1207.
arXiv:2405.02460.

[65] J. T. Mackereth, J. Bovy, H. W. Leung, R. P. Schiavon, W. H. Trick,

W. J. Chaplin, K. Cunha, D. K. Feuillet, S. R. Majewski, M. Martig,
A. Miglio, D. Nidever, M. H. Pinsonneault, V. S. Aguirre, J. Sobeck,
J. Tayar, G. Zasowski, Dynamical heating across the Milky Way disc
using APOGEE and Gaia, Mon. Not. R. Astron. Soc. 489 (2019) 176–
195. doi:10.1093/mnras/stz1521. arXiv:1901.04502.

[66] S. R. Kulkarni, C. Heiles, Neutral hydrogen and the diffuse interstellar

medium., in: K. I. Kellermann, G. L. Verschuur (Eds.), Galactic and
Extragalactic Radio Astronomy, 1988, pp. 95–153.

[67] J. M. Dickey, F. J. Lockman, H I in the galaxy., Annu. Rev. Astron.

Astrophys. 28 (1990) 215–261. doi:10.1146/annurev.aa.28.090190.
001243.

[68] P. M. W. Kalberla, J. Kerp, The Hi Distribution of the Milky

Way, Annu. Rev. Astron. Astrophys. 47 (2009) 27–61. doi:10.1146/
annurev-astro-082708-101823.

[69] N. M. McClure-Griffiths, S. Stanimirović, D. R. Rybarczyk, Atomic

Hydrogen in the Milky Way: A Stepping Stone in the Evolution of
Galaxies, Annu. Rev. Astron. Astrophys. 61 (2023) 19–63. doi:10.
1146/annurev-astro-052920-104851. arXiv:2307.08464.

[70] N. Z. Scoville, D. B. Sanders, H2 in the Galaxy, in: D. J. Hollenbach,

J. Thronson, Harley A. (Eds.), Interstellar Processes, volume 134, 1987,
p. 21. doi:10.1007/978-94-009-3861-8_2.

[71] M. Heyer, T. M. Dame, Molecular Clouds in the Milky Way,

Annu. Rev. Astron. Astrophys. 53 (2015) 583–629. doi:10.1146/
annurev-astro-082214-122324.

184
[72] J. S. Young, N. Z. Scoville, Molecular gas in galaxies., Annu. Rev.
Astron. Astrophys. 29 (1991) 581–625. doi:10.1146/annurev.aa.29.
090191.003053.

[73] A. Omont, Molecules in galaxies, Reports on Progress in

Physics 70 (2007) 1099–1176. doi:10.1088/0034-4885/70/7/R03.
arXiv:0709.3814.

[74] C. F. McKee, J. P. Ostriker, A theory of the interstellar medium: three

components regulated by supernova explosions in an inhomogeneous
substrate., Astrophys. J. 218 (1977) 148–169. doi:10.1086/155667.

[75] S. R. Kulkarni, C. Heiles, The Atomic Component, in: D. J. Hollen-

bach, H. A. Thronson, Jr. (Eds.), Interstellar Processes, volume 134,
1987, p. 87. doi:10.1007/978-94-009-3861-8_5.

[76] J. M. Dickey, S. Strasser, B. M. Gaensler, M. Haverkorn, D. Kavars,

N. M. McClure-Griffiths, J. Stil, A. R. Taylor, The Outer Disk of
the Milky Way Seen in λ21 cm Absorption, Astrophys. J. 693 (2009)
1250–1260. doi:10.1088/0004-637X/693/2/1250. arXiv:0901.0968.

[77] F. J. Lockman, C. S. Gehman, Vertical Distribution and Support of

Galactic H i, Astrophys. J. 382 (1991) 182. doi:10.1086/170706.

[78] C. A. Narayan, K. Saha, C. J. Jog, Constraints on the halo

density profile using HI flaring in the outer Galaxy, Astron.
Astrophys. 440 (2005) 523–530. doi:10.1051/0004-6361:20041055.
arXiv:astro-ph/0508358.

[79] F. Lelli, S. S. McGaugh, J. M. Schombert, SPARC: Mass Models for 175

Disk Galaxies with Spitzer Photometry and Accurate Rotation Curves,
The Astronomical Journal 152 (2016) 157. doi:10.3847/0004-6256/
152/6/157. arXiv:1606.09251.

[80] P. Biswas, V. Kalinova, N. Roy, N. N. Patra, N. Tyulneva, The GMRT

archive atomic gas survey - II. Mass modelling and dark matter halo
properties across late-type spirals, Mon. Not. R. Astron. Soc. 524
(2023) 6213–6228. doi:10.1093/mnras/stad2285. arXiv:2307.13738.

185
[81] J. N. Bahcall, R. M. Soneira, The universe at faint magnitudes. I. Mod-
els for the Galaxy and the predicted star counts., The Astrophysical
Journal Supplement 44 (1980) 73–110. doi:10.1086/190685.

[82] D. Mera, G. Chabrier, R. Schaeffer, Towards a consistent model

of the Galaxy. II. Derivation of the model, Astron. Astro-
phys. 330 (1998) 953–962. doi:10.48550/arXiv.astro-ph/9801050.
arXiv:astro-ph/9801050.

[83] F. Sylos Labini, Ž. Chrobáková, R. Capuzzo-Dolcetta, M. López-

Corredoira, Mass Models of the Milky Way and Estimation of Its
Mass from the Gaia DR3 Data Set, Astrophys. J. 945 (2023) 3.
doi:10.3847/1538-4357/acb92c. arXiv:2302.01379.

[84] V. C. Rubin, W. K. Ford, Jr., N. Thonnard, Rotational properties of

21 SC galaxies with a large range of luminosities and radii, from NGC
4605 (R=4kpc) to UGC 2885 (R=122kpc)., Astrophys. J. 238 (1980)
471–487. doi:10.1086/158003.

[85] V. C. Rubin, The Rotation of Spiral Galaxies, Science 220 (1983)

1339–1344. doi:10.1126/science.220.4604.1339.

[86] D. Mihalas, P. M. Routly, Galactic astronomy, 1968.

[87] K. Rohlfs, Lectures on density wave theory, volume 69, 1977. doi:10.
1007/3-540-08448-7.

[88] J. H. Oort, The force exerted by the stellar system in the direction
perpendicular to the galactic plane and some related problems, B.A.N.
6 (1932) 249.

[89] J. N. Bahcall, The distribution of stars perpendicular to galactic disk,

Astrophys. J. 276 (1984) 156–168. doi:10.1086/161600.

[90] S. Sarkar, C. J. Jog, Gravitational potential energy of a multi-

component galactic disk, Astron. Astrophys. 665 (2022) A23. doi:10.
1051/0004-6361/202243184. arXiv:2207.09393.

[91] S. Sarkar, C. J. Jog, Flaring stellar disk in the low surface brightness
galaxy UGC 7321, Astron. Astrophys. 628 (2019) A58. doi:10.1051/
0004-6361/201935430. arXiv:1905.02735.

186
 [92] V. Trimble, Existence and nature of dark matter in the universe., Annu.
Rev. Astron. Astrophys. 25 (1987) 425–472. doi:10.1146/annurev.aa.
25.090187.002233.

[93] J. N. Bahcall, Self-consistent determinations of the total amount of

matter near the sun., Astrophys. J. 276 (1984) 169–181. doi:10.1086/
161601.

[94] C. Flynn, B. Fuchs, Density of dark matter in the Galactic disk, Mon.
Not. R. Astron. Soc. 270 (1994) 471–479. doi:10.1093/mnras/270.3.
471.

[95] J. Holmberg, C. Flynn, The local density of matter mapped by Hip-

parcos, Mon. Not. R. Astron. Soc. 313 (2000) 209–216. doi:10.1046/
j.1365-8711.2000.02905.x. arXiv:astro-ph/9812404.

[96] O. Bienaymé, B. Famaey, A. Siebert, K. C. Freeman, B. K. Gibson,

G. Gilmore, E. K. Grebel, J. Bland-Hawthorn, G. Kordopatis, U. Mu-
nari, J. F. Navarro, Q. Parker, W. Reid, G. M. Seabroke, A. Siviero,
M. Steinmetz, F. Watson, R. F. G. Wyse, T. Zwitter, Weighing the
local dark matter with RAVE red clump stars, Astron. Astrophys. 571
(2014) A92. doi:10.1051/0004-6361/201424478. arXiv:1406.6896.

[97] C. F. McKee, A. Parravano, D. J. Hollenbach, Stars, Gas, and Dark

Matter in the Solar Neighborhood, Astrophys. J. 814 (2015) 13. doi:10.
1088/0004-637X/814/1/13. arXiv:1509.05334.

[98] C. Moni Bidin, G. Carraro, R. A. Méndez, R. Smith, Kinematical and

Chemical Vertical Structure of the Galactic Thick Disk. II. A Lack of
Dark Matter in the Solar Neighborhood, Astrophys. J. 751 (2012) 30.
doi:10.1088/0004-637X/751/1/30. arXiv:1204.3924.

[99] J. Bovy, S. Tremaine, On the Local Dark Matter Density, Astrophys. J.

756 (2012) 89. doi:10.1088/0004-637X/756/1/89. arXiv:1205.4033.

[100] J. H. J. Hagen, A. Helmi, The vertical force in the solar neighbourhood

using red clump stars in TGAS and RAVE. Constraints on the local
dark matter density, Astron. Astrophys. 615 (2018) A99. doi:10.1051/
0004-6361/201832903. arXiv:1802.09291.

187
[101] J. Buch, J. S. C. Leung, J. Fan, Using Gaia DR2 to constrain local
dark matter density and thin dark disk, Journal of Cosmology and
Astrophysics 2019 (2019) 026. doi:10.1088/1475-7516/2019/04/026.
arXiv:1808.05603.

[102] R. Guo, C. Liu, S. Mao, X.-X. Xue, R. J. Long, L. Zhang, Measuring

the local dark matter density with LAMOST DR5 and Gaia DR2,
Mon. Not. R. Astron. Soc. 495 (2020) 4828–4844. doi:10.1093/mnras/
staa1483. arXiv:2005.12018.

[103] M. S. Nitschai, A.-C. Eilers, N. Neumayer, M. Cappellari, H.-W.

Rix, Dynamical Model of the Milky Way Using APOGEE and Gaia
Data, Astrophys. J. 916 (2021) 112. doi:10.3847/1538-4357/ac04b5.
arXiv:2106.05286.

[104] J. I. Read, The local dark matter density, Journal of Physics

G Nuclear Physics 41 (2014) 063101. doi:10.1088/0954-3899/41/6/
063101. arXiv:1404.1938.

[105] P. F. de Salas, A. Widmark, Dark matter local density de-

termination: recent observations and future prospects, Reports
on Progress in Physics 84 (2021) 104901. doi:10.1088/1361-6633/
ac24e7. arXiv:2012.11477.

[106] L. E. Strigari, Galactic searches for dark matter, Physics

Reports 531 (2013) 1–88. doi:10.1016/j.physrep.2013.05.004.
arXiv:1211.7090.

[107] J. Binney, M. Merrifield, Galactic Astronomy, 1998.

[108] F. J. Lockman, The HI halo in the inner galaxy., Astrophys. J. 283

(1984) 90–97. doi:10.1086/162277.

[109] J. H. Oort, Radio Data on the Distribution and Motion of Interstellar

Gas, in: L. Woltjer (Ed.), The Distribution and Motion of Interstellar
Matter in Galaxies, 1962, p. 3.

[110] P. M. W. Kalberla, L. Dedes, Global properties of the H I distribu-

tion in the outer Milky Way. Planar and extra-planar gas, Astron.
Astrophys. 487 (2008) 951–963. doi:10.1051/0004-6361:20079240.
arXiv:0804.4831.

188
[111] E. S. Levine, L. Blitz, C. Heiles, The Vertical Structure of the Outer
Milky Way H I Disk, Astrophys. J. 643 (2006) 881–896. doi:10.1086/
503091. arXiv:astro-ph/0601697.

[112] C. J. Jog, C. A. Narayan, Vertical distribution of Galactic disc stars and

gas constrained by a molecular cloud complex, Mon. Not. R. Astron.
Soc. 327 (2001) 1021–1028. doi:10.1046/j.1365-8711.2001.04780.x.
arXiv:astro-ph/0110331.

[113] D. B. Sanders, D. P. Clemens, N. Z. Scoville, P. M. Solomon, Molecular

Cloud Clusters and Chains, in: H. van Woerden, R. J. Allen, W. B.
Burton (Eds.), The Milky Way Galaxy, volume 106 of IAU Symposium,
1985, p. 329.

[114] A. R. Rivolo, P. M. Solomon, D. B. Sanders, The Massachusetts–Stony

Brook Galactic Plane CO Survey: Statistical Clustering Properties of
Molecular Cloud Cores, Astrophys. J. Lett. 301 (1986) L19. doi:10.
1086/184615.

[115] T. M. Dame, H. Ungerechts, R. S. Cohen, E. J. de Geus, I. A. Grenier,

J. May, D. C. Murphy, L. A. Nyman, P. Thaddeus, A Composite
CO Survey of the Entire Milky Way, Astrophys. J. 322 (1987) 706.
doi:10.1086/165766.

[116] M. Schmidt, A model of the distribution of mass in the Galactic Sys-

tem, B.A.N. 13 (1956) 15.

[117] W. H. Press, B. P. Flannery, S. A. Teukolsky, Numerical recipes. The

art of scientific computing, 1986.

[118] M. Nandakumar, N. Roy, C. J. Jog, K. M. Menten, Observa-

tional evidence for local vertical constraining of H I by molecular
cloud complexes, Mon. Not. R. Astron. Soc. 528 (2024) 3992–4009.
doi:10.1093/mnras/stae212. arXiv:2401.08168.

[119] HI4PI Collaboration, N. Ben Bekhti, L. Flöer, R. Keller, J. Kerp,

D. Lenz, B. Winkel, J. Bailin, M. R. Calabretta, L. Dedes, H. A.
Ford, B. K. Gibson, U. Haud, S. Janowiecki, P. M. W. Kalberla,
F. J. Lockman, N. M. McClure-Griffiths, T. Murphy, H. Nakan-
ishi, D. J. Pisano, L. Staveley-Smith, HI4PI: A full-sky H I survey

189
 based on EBHIS and GASS, Astron. Astrophys. 594 (2016) A116.
doi:10.1051/0004-6361/201629178. arXiv:1610.06175.

[120] R. P. Olling, On the Usage of Flaring Gas Layers to Determine the

Shape of Dark Matter Halos, The Astronomical Journal 110 (1995)
591. doi:10.1086/117545. arXiv:astro-ph/9505002.

[121] C. J. Jog, P. M. Solomon, Two-fluid gravitational instabilities in a

galactic disk, Astrophys. J. 276 (1984) 114–126. doi:10.1086/161597.

[122] S. Sharma, M. R. Hayden, J. Bland-Hawthorn, D. Stello, S. Buder,

J. C. Zinn, T. Kallinger, M. Asplund, G. M. De Silva, V. D’Orazi,
K. Freeman, J. Kos, G. F. Lewis, J. Lin, K. Lind, S. Martell, J. D.
Simpson, R. A. Wittenmyer, D. B. Zucker, T. Zwitter, B. Chen, K. Co-
tar, J. Esdaile, M. Hon, J. Horner, D. Huber, P. R. Kafle, S. Khanna,
Y.-S. Ting, D. M. Nataf, T. Nordlander, M. H. M. Saadon, T. Tepper-
Garcia, C. G. Tinney, G. Traven, F. Watson, D. Wright, R. F. G.
Wyse, Fundamental relations for the velocity dispersion of stars in
the Milky Way, Mon. Not. R. Astron. Soc. 506 (2021) 1761–1776.
doi:10.1093/mnras/stab1086. arXiv:2004.06556.

[123] B. M. Lewis, Face-on galaxies., Astrophys. J. 285 (1984) 453–457.

doi:10.1086/162474.

[124] P. Kamphuis, The Structure and Kinematics of Halos in Disk Galaxies,

Ph.D. thesis, University of Groningen, Netherlands, 2008.

[125] J. M. Dickey, The Z Velocity Dispersion of Outer HI Disks, in: L. Blitz,

P. J. Teuben (Eds.), Unsolved Problems of the Milky Way, volume 169
of IAU Symposium, 1996, p. 489.

[126] S. M. Kent, T. M. Dame, G. Fazio, Galactic Structure from the Space-

lab Infrared Telescope. II. Luminosity Models of the Milky Way, As-
trophys. J. 378 (1991) 131. doi:10.1086/170413.

[127] I. Minchev, M. Martig, D. Streich, C. Scannapieco, R. S. de Jong,

M. Steinmetz, On the Formation of Galactic Thick Disks, As-
trophys. J. Lett. 804 (2015) L9. doi:10.1088/2041-8205/804/1/L9.
arXiv:1502.06606.

190
[128] P. M. W. Kalberla, J. Kerp, L. Dedes, U. Haud, Does the Stellar
Distribution Flare? A Comparison of Stellar Scale Heights with LAB
H I Data, Astrophys. J. 794 (2014) 90. doi:10.1088/0004-637X/794/
1/90. arXiv:1408.5334.

[129] D. Sotillo-Ramos, M. Donnari, A. Pillepich, N. Frankel, D. Nelson,

V. Springel, L. Hernquist, Disc flaring with TNG50: diversity across
Milky Way and M31 analogues, Mon. Not. R. Astron. Soc. 523 (2023)
3915–3938. doi:10.1093/mnras/stad1485. arXiv:2303.16228.

[130] O. V. Abramova, A. V. Zasov, The gas density and “volume” Schmidt

law for spiral galaxies, Astronomy Reports 52 (2008) 257–269. doi:10.
1134/S106377290804001X.

[131] N. N. Patra, Molecular scale height in spiral galaxies, Mon.

Not. R. Astron. Soc. 484 (2019) 81–92. doi:10.1093/mnras/sty3493.
arXiv:2004.13053.

[132] R. A. Swaters, R. Sancisi, J. M. van der Hulst, The H I Halo of

NGC 891, Astrophys. J. 491 (1997) 140–145. doi:10.1086/304958.
arXiv:astro-ph/9707150.

[133] F. Fraternali, G. van Moorsel, R. Sancisi, T. Oosterloo, Deep H I

Survey of the Spiral Galaxy NGC 2403, The Astronomical Journal 123
(2002) 3124–3140. doi:10.1086/340358. arXiv:astro-ph/0203405.

[134] L. D. Matthews, K. Wood, High-Latitude H I in the Low Surface

Brightness Galaxy UGC 7321, Astrophys. J. 593 (2003) 721–732.
doi:10.1086/376602. arXiv:astro-ph/0305402.

[135] F. Fraternali, T. A. Oosterloo, R. Sancisi, R. Swaters, The Extra-planar

Neutral Gas in the Edge-on Spiral Galaxy NGC 891, in: R. Braun
(Ed.), Extra-Planar Gas, volume 331 of Astronomical Society of the Pa-
cific Conference Series, 2005, p. 239. doi:10.48550/arXiv.astro-ph/
0410375. arXiv:astro-ph/0410375.

[136] R. Boomsma, T. A. Oosterloo, F. Fraternali, J. M. van der Hulst,

R. Sancisi, HI holes and high-velocity clouds in the spiral galaxy
NGC 6946, Astron. Astrophys. 490 (2008) 555–570. doi:10.1051/
0004-6361:200810120. arXiv:0807.3339.

191
[137] P. Kamphuis, R. F. Peletier, P. C. van der Kruit, T. A. Oosterloo,
R. Sancisi, A Study of Extra-Planar HI Gas, in: R. S. DE JONG
(Ed.), Island Universes, volume 3 of Astrophysics and Space Science
Proceedings, 2007, p. 303. doi:10.1007/978-1-4020-5573-7_51.

[138] T. de Zeeuw, D. Pfenniger, Potential-density pairs for galaxies., Mon.

Not. R. Astron. Soc. 235 (1988) 949–995. doi:10.1093/mnras/235.3.
949.

[139] J. F. Navarro, C. S. Frenk, S. D. M. White, The Structure of Cold

Dark Matter Halos, Astrophys. J. 462 (1996) 563. doi:10.1086/177173.
arXiv:astro-ph/9508025.

[140] A. Banerjee, C. J. Jog, Progressively More Prolate Dark Matter Halo

in the Outer Galaxy as Traced by Flaring H I Gas, Astrophys. J. Lett.
732 (2011) L8. doi:10.1088/2041-8205/732/1/L8. arXiv:1103.5821.

[141] A. Banerjee, C. J. Jog, E. Brinks, I. Bagetakos, Theoretical determina-

tion of H I vertical scale heights in the dwarf galaxies DDO 154, Ho II,
IC 2574 and NGC 2366, Mon. Not. R. Astron. Soc. 415 (2011) 687–694.
doi:10.1111/j.1365-2966.2011.18745.x. arXiv:1103.4494.

[142] M. L. Mateo, Dwarf Galaxies of the Local Group, Annu. Rev. As-
tron. Astrophys. 36 (1998) 435–506. doi:10.1146/annurev.astro.36.
1.435. arXiv:astro-ph/9810070.

[143] E. Brinks, F. Walter, J. Ott, Bloated Dwarfs: The Thickness of the HI

Disks in Irregular Galaxies, in: E. Athanassoula, A. Bosma, R. Mujica
(Eds.), Disks of Galaxies: Kinematics, Dynamics and Peturbations,
volume 275 of Astronomical Society of the Pacific Conference Series,
2002, pp. 57–60.

[144] R. Sancisi, R. J. Allen, Neutral hydrogen observations of the edge-on

disk galaxy NGC 891., Astron. Astrophys. 74 (1979) 73–84.

[145] R. H. Warmels, The HI properties of spiral galaxies in the Virgo clus-

ter. III. The HI surface density distribution in 36 galaxies., Astron.
Astrophys. Supplement 72 (1988) 427–447.

[146] J. Kormendy, K. C. Freeman, Scaling Laws for Dark Matter Halos in

Late-Type and Dwarf Spheroidal Galaxies, in: S. Ryder, D. Pisano,

192
 M. Walker, K. Freeman (Eds.), Dark Matter in Galaxies, volume
220 of IAU Symposium, 2004, p. 377. doi:10.48550/arXiv.astro-ph/
0407321. arXiv:astro-ph/0407321.

[147] A. Banerjee, C. J. Jog, Why are some galaxy discs extremely thin?,
Mon. Not. R. Astron. Soc. 431 (2013) 582–588. doi:10.1093/mnras/
stt186. arXiv:1210.8244.

[148] L. D. Matthews, The Extraordinary “Superthin” Spiral Galaxy UGC

7321. II. The Vertical Disk Structure, The Astronomical Journal 120
(2000) 1764–1778. doi:10.1086/301555. arXiv:astro-ph/0006412.

[149] L. D. Matthews, J. S. Gallagher, III, W. van Driel, The Extraor-

dinary “Superthin” Spiral Galaxy UGC 7321. I. Disk Color Gradi-
ents and Global Properties from Multiwavelength Observations, The
Astronomical Journal 118 (1999) 2751–2766. doi:10.1086/301128.
arXiv:astro-ph/9909142.

[150] J. M. Uson, L. D. Matthews, H I Imaging Observations of Superthin

Galaxies. I. UGC 7321, The Astronomical Journal 125 (2003) 2455–
2472. doi:10.1086/374627. arXiv:astro-ph/0303275.

[151] S. Kurapati, A. Banerjee, J. N. Chengalur, D. Makarov, S. Borisov,

A. Afanasiev, A. Antipova, Mass modelling of a superthin galaxy,
FGC 1540, Mon. Not. R. Astron. Soc. 479 (2018) 5686–5695. doi:10.
1093/mnras/sty1856. arXiv:1807.02992.

[152] F. J. Sánchez-Salcedo, C. Flynn, J. A. de Diego, Jeans Analy-

sis of the Galactic Thick Disk and the Local Dark Matter Den-
sity, Astrophys. J. 817 (2016) 13. doi:10.3847/0004-637X/817/1/13.
arXiv:1511.08199.

[153] C. Wegg, O. Gerhard, M. Bieth, The gravitational force field of the

Galaxy measured from the kinematics of RR Lyrae in Gaia, Mon. Not.
R. Astron. Soc. 485 (2019) 3296–3316. doi:10.1093/mnras/stz572.
arXiv:1806.09635.

[154] C. L. Perry, The galactic force law K(z)., The Astronomical Journal
74 (1969) 139–165. doi:10.1086/110787.

193
[155] Y. Jing, C. Du, J. Gu, Y. Jia, X. Peng, Y. Chen, Z. Wu, J. Ma, X. Zhou,
Z. Cao, Y. Hou, Y. Wang, Y. Zhang, Kinematics of the Galactic disc
from a LAMOST dwarf sample, Mon. Not. R. Astron. Soc. 463 (2016)
3390–3397. doi:10.1093/mnras/stw2230. arXiv:1609.00455.

[156] Q. Xia, C. Liu, S. Mao, Y. Song, L. Zhang, R. J. Long, Y. Zhang,

Y. Hou, Y. Wang, Y. Wu, Determining the local dark matter density
with LAMOST data, Mon. Not. R. Astron. Soc. 458 (2016) 3839–3850.
doi:10.1093/mnras/stw565. arXiv:1510.06810.

[157] W. X. Sun, Y. Huang, H. F. Wang, C. Wang, M. Zhang, X. Y. Li, B. Q.

Chen, H. W. Zhang, D. D. Wei, D. K. Jiang, X. W. Liu, Mapping the
Galactic Disk with the LAMOST and Gaia Red Clump Sample. V. On
the Origin of the “Young” [α/Fe]-enhanced Stars, Astrophys. J. 903
(2020) 12. doi:10.3847/1538-4357/abb1b7. arXiv:2008.10218.

[158] S. Garbari, J. I. Read, G. Lake, Limits on the local dark matter den-
sity, Mon. Not. R. Astron. Soc. 416 (2011) 2318–2340. doi:10.1111/
j.1365-2966.2011.19206.x. arXiv:1105.6339.

[159] P. Yoachim, J. J. Dalcanton, Structural Parameters of Thin and Thick

Disks in Edge-on Disk Galaxies, The Astronomical Journal 131 (2006)
226–249. doi:10.1086/497970. arXiv:astro-ph/0508460.

[160] S. Comerón, B. G. Elmegreen, J. H. Knapen, H. Salo, E. Lau-

rikainen, J. Laine, E. Athanassoula, A. Bosma, K. Sheth, M. W. Re-
gan, J. L. Hinz, A. Gil de Paz, K. Menéndez-Delmestre, T. Mizu-
sawa, J.-C. Muñoz-Mateos, M. Seibert, T. Kim, D. M. Elmegreen,
D. A. Gadotti, L. C. Ho, B. W. Holwerda, J. Lappalainen, E. Schin-
nerer, R. Skibba, Thick Disks of Edge-on Galaxies Seen through the
Spitzer Survey of Stellar Structure in Galaxies (S4 G): Lair of Missing
Baryons?, Astrophys. J. 741 (2011) 28. doi:10.1088/0004-637X/741/
1/28. arXiv:1108.0037.

[161] N. N. Patra, H I scale height in spiral galaxies, Mon. Not. R. Astron.

Soc. 499 (2020) 2063–2075. doi:10.1093/mnras/staa2959.

[162] X. Li, Y. Shi, Z.-Y. Zhang, J. Chen, X. Yu, J. Wang, Q. Gu, S. Li,
The H I gas disc thickness of the ultra-diffuse galaxy AGC 242019,

194
 Mon. Not. R. Astron. Soc. 516 (2022) 4220–4227. doi:10.1093/mnras/
stac2340. arXiv:2208.12495.

[163] H.-H. Wang, R. S. Klessen, C. P. Dullemond, F. C. van den Bosch,

B. Fuchs, Equilibrium initialization and stability of three-dimensional
gas discs, Mon. Not. R. Astron. Soc. 407 (2010) 705–720. doi:10.1111/
j.1365-2966.2010.16942.x. arXiv:1004.5593.

[164] M. Schmidt, The Rate of Star Formation., Astrophys. J. 129 (1959)

243. doi:10.1086/146614.

[165] E. C. Ostriker, C.-G. Kim, Pressure-regulated, Feedback-modulated

Star Formation in Disk Galaxies, Astrophys. J. 936 (2022) 137. doi:10.
3847/1538-4357/ac7de2. arXiv:2206.00681.

[166] S. Khrapov, A. Khoperskov, V. Korchagin, Modeling of Spiral Struc-

ture in a Multi-Component Milky Way-Like Galaxy, Galaxies 9 (2021)
29. doi:10.3390/galaxies9020029. arXiv:2105.03198.

[167] C. Bacchini, F. Fraternali, G. Iorio, G. Pezzulli, Volumetric star for-

mation laws of disc galaxies, Astron. Astrophys. 622 (2019) A64.
doi:10.1051/0004-6361/201834382. arXiv:1810.03616.

[168] J. Forbes, M. Krumholz, A. Burkert, Evolving Gravitationally Un-

stable Disks over Cosmic Time: Implications for Thick Disk Forma-
tion, Astrophys. J. 754 (2012) 48. doi:10.1088/0004-637X/754/1/48.
arXiv:1112.1410.

[169] S. Aniyan, K. C. Freeman, M. Arnaboldi, O. E. Gerhard, L. Coc-

cato, M. Fabricius, K. Kuijken, M. Merrifield, A. A. Ponomareva, Re-
solving the disc-halo degeneracy - I: a look at NGC 628, Mon. Not.
R. Astron. Soc. 476 (2018) 1909–1930. doi:10.1093/mnras/sty310.
arXiv:1802.00465.

[170] T. Tsukui, E. Wisnioski, J. Bland-Hawthorn, K. Freeman, The emer-

gence of galactic thin and thick discs across cosmic history, arXiv
e-prints (2024) arXiv:2409.15909. doi:10.48550/arXiv.2409.15909.
arXiv:2409.15909.

[171] J. Binney, S. Tremaine, Galactic Dynamics: Second Edition, 2008.

195
[172] S. Chandrasekhar, Ellipsoidal figures of equilibrium, 1969.
[173] G. L. Camm, The virial theorem for statistical distributions of stars,
in: Les Nouvelles Méthodes de la Dynamique Stellaire, 1967, p. 141.
[174] D. Tamburro, H. W. Rix, A. K. Leroy, M. M. Mac Low, F. Walter,
R. C. Kennicutt, E. Brinks, W. J. G. de Blok, What is Driving the H I
Velocity Dispersion?, The Astronomical Journal 137 (2009) 4424–4435.
doi:10.1088/0004-6256/137/5/4424. arXiv:0903.0183.
[175] W. Sun, Y. Huang, H. Shen, C. Wang, H. Zhang, Z. Tian, X. Liu,
B. Jiang, Mapping the Galactic Disk with the LAMOST and Gaia Red
Clump Sample. VIII. Mapping the Kinematics of the Galactic Disk Us-
ing Mono-age and Mono-abundance Stellar Populations, Astrophys. J.
961 (2024) 141. doi:10.3847/1538-4357/ad06ad. arXiv:2310.15408.
[176] N. Uppal, S. Ganesh, M. Schultheis, Warp and flare of the old Galactic
disc as traced by the red clump stars, Mon. Not. R. Astron. Soc. 527
(2024) 4863–4873. doi:10.1093/mnras/stad3525. arXiv:2311.09616.
[177] Y. Yu, H.-F. Wang, W.-Y. Cui, L.-L. Li, C. Liu, B. Zhang, H. Tian,
Z.-Y. Huo, J. Ju, Z.-C. Liu, F. Wen, S. Feng, The Flare and
Warp of the Young Stellar Disk Traced with LAMOST DR5 OB-type
Stars, Astrophys. J. 922 (2021) 80. doi:10.3847/1538-4357/ac1e91.
arXiv:2102.00731.
[178] Ž. Chrobáková, R. Nagy, M. López-Corredoira, Warp and flare of
the Galactic disc revealed with supergiants by Gaia EDR3, As-
tron. Astrophys. 664 (2022) A58. doi:10.1051/0004-6361/202243296.
arXiv:2206.08230.
[179] I. Ciucǎ, D. Kawata, J. Lin, L. Casagrande, G. Seabroke, M. Crop-
per, The vertical metallicity gradients of mono-age stellar popula-
tions in the Milky Way with the RAVE and Gaia data, Mon. Not.
R. Astron. Soc. 475 (2018) 1203–1212. doi:10.1093/mnras/stx3285.
arXiv:1706.05005.
[180] J. Bovy, A. Bahmanyar, T. K. Fritz, N. Kallivayalil, The Shape of the
Inner Milky Way Halo from Observations of the Pal 5 and GD–1 Stellar
Streams, Astrophys. J. 833 (2016) 31. doi:10.3847/1538-4357/833/
1/31. arXiv:1609.01298.

196
[181] J. T. Mackereth, J. Bovy, R. P. Schiavon, G. Zasowski, K. Cunha,
P. M. Frinchaboy, A. E. Garcı́a Perez, M. R. Hayden, J. Holtzman,
S. R. Majewski, S. Mészáros, D. L. Nidever, M. Pinsonneault, M. D.
Shetrone, The age-metallicity structure of the Milky Way disc using
APOGEE, Mon. Not. R. Astron. Soc. 471 (2017) 3057–3078. doi:10.
1093/mnras/stx1774. arXiv:1706.00018.

[182] A. V. Kasparova, I. Y. Katkov, I. V. Chilingarian, O. K. Silchenko,

A. V. Moiseev, S. B. Borisov, The diversity of thick galactic discs,
Mon. Not. R. Astron. Soc. 460 (2016) L89–L93. doi:10.1093/mnrasl/
slw083. arXiv:1604.07624.

[183] R. M. Rich, A. Mosenkov, H. Lee-Saunders, A. Koch, J. Kormendy,

J. Kennefick, N. Brosch, L. Sales, J. Bullock, A. Burkert, M. Collins,
M. Cooper, M. Fusco, D. Reitzel, D. Thilker, D. G. Milewski, L. Elias,
M. L. Saade, L. De Groot, The haloes and environments of nearby
galaxies (HERON) - I. Imaging, sample characteristics, and envelope
diameters, Mon. Not. R. Astron. Soc. 490 (2019) 1539–1569. doi:10.
1093/mnras/stz2106. arXiv:1907.10706.

[184] L. Ossa-Fuentes, A. S. Borlaff, J. E. Beckman, P. M. Marcum, M. N.

Fanelli, Flares, Warps, Truncations, and Satellite: The Ultra-thin
Galaxy UGC 11859, Astrophys. J. 951 (2023) 149. doi:10.3847/
1538-4357/acd54c. arXiv:2307.09521.

[185] N. Ranaivoharimina, T. Randriamampandry, J. Wang, K. Menéndez-

Delmestre, T. S. Gonçalves, On the Stellar Disk Vertical Scale Height
of Edge-on Galaxies from S4 G, Astrophys. J. 977 (2024) 66. doi:10.
3847/1538-4357/ad85d5. arXiv:2410.09762.

[186] D. Kawata, R. J. J. Grand, B. K. Gibson, L. Casagrande, J. A. S. Hunt,

C. B. Brook, Impacts of a flaring star-forming disc and stellar radial
mixing on the vertical metallicity gradient, Mon. Not. R. Astron. Soc.
464 (2017) 702–712. doi:10.1093/mnras/stw2363. arXiv:1604.07412.

[187] T. Haines, E. D’Onghia, B. Famaey, C. Laporte, L. Hernquist, Im-

plications of a Time-varying Galactic Potential for Determinations of
the Dynamical Surface Density, Astrophys. J. Lett. 879 (2019) L15.
doi:10.3847/2041-8213/ab25f3. arXiv:1903.00607.

197
[188] E. D’Onghia, Disk-stability Constraints on the Number of Arms in
Spiral Galaxies, Astrophys. J. Lett. 808 (2015) L8. doi:10.1088/
2041-8205/808/1/L8. arXiv:1507.00724.

[189] W.-T. Kim, E. C. Ostriker, Gravitational Runaway and Turbulence

Driving in Star-Gas Galactic Disks, Astrophys. J. 660 (2007) 1232–
1245. doi:10.1086/513176. arXiv:astro-ph/0701755.

[190] M. Behrendt, A. Burkert, M. Schartmann, Structure formation in gas-

rich galactic discs with finite thickness: from discs to rings, Mon. Not.
R. Astron. Soc. 448 (2015) 1007–1019. doi:10.1093/mnras/stv027.
arXiv:1408.5902.

[191] S. Ghosh, C. J. Jog, Effect of finite disk-thickness on swing amplifica-

tion of non-axisymmetric perturbations in a sheared galactic disk, As-
tron. Astrophys. 617 (2018) A47. doi:10.1051/0004-6361/201832988.
arXiv:1806.01439.

[192] S. Ghosh, C. J. Jog, Dynamical interplay of disc thickness and interstel-

lar gas: Implication for the longevity of spiral density waves, Astron.
Astrophys. 658 (2022) A171. doi:10.1051/0004-6361/202142174.
arXiv:2111.10893.

[193] P. Natarajan (Ed.), The Shapes of Galaxies and Their Dark Halos,
2002.

[194] R. H. Wechsler, J. L. Tinker, The Connection Between Galax-

ies and Their Dark Matter Halos, Annu. Rev. Astron. Astrophys.
56 (2018) 435–487. doi:10.1146/annurev-astro-081817-051756.
arXiv:1804.03097.

[195] F. Combes, Properties of dark matter haloes, New Astronomy

Review 46 (2002) 755–766. doi:10.1016/S1387-6473(02)00244-0.
arXiv:astro-ph/0206126.

[196] V. C. Rubin, W. K. Ford, Jr., N. Thonnard, Extended rotation curves

of high-luminosity spiral galaxies. IV. Systematic dynamical properties,
Sa -¿ Sc., Astrophys. J. Lett. 225 (1978) L107–L111. doi:10.1086/
182804.

198
[197] M. S. Roberts, R. N. Whitehurst, The rotation curve and geometry
of M31 at large galactocentric distances., Astrophys. J. 201 (1975)
327–346. doi:10.1086/153889.

[198] A. Bosma, The distribution and kinematics of neutral hydrogen in spi-

ral galaxies of various morphological types, Ph.D. thesis, University of
Groningen, Netherlands, 1978.

[199] T. S. van Albada, J. N. Bahcall, K. Begeman, R. Sancisi, Distribution

of dark matter in the spiral galaxy NGC 3198., Astrophys. J. 295
(1985) 305–313. doi:10.1086/163375.

[200] S. M. Kent, Dark matter in spiral galaxies. I. Galaxies with opti-

cal rotation curves., The Astronomical Journal 91 (1986) 1301–1327.
doi:10.1086/114106.

[201] A. H. Broeils, S. Courteau, Modelling the Mass Distribution in Spiral

Galaxies, in: M. Persic, P. Salucci (Eds.), Dark and Visible Matter in
Galaxies and Cosmological Implications, volume 117 of Astronomical
Society of the Pacific Conference Series, 1997, p. 74. doi:10.48550/
arXiv.astro-ph/9610264. arXiv:astro-ph/9610264.

[202] K. M. Ashman, Dark Matter in Galaxies, Publications of the Astro-

nomical Society of the Pacific 104 (1992) 1109. doi:10.1086/133099.

[203] P. D. Sackett, L. S. Sparke, The Dark Halo of the Polar-Ring Galaxy

NGC 4650A, Astrophys. J. 361 (1990) 408. doi:10.1086/169206.

[204] R. Giovanelli, M. P. Haynes, Extragalactic neutral hydrogen., in: K. I.

Kellermann, G. L. Verschuur (Eds.), Galactic and Extragalactic Radio
Astronomy, 1988, pp. 522–562.

[205] C. A. Narayan, C. J. Jog, The puzzle about the radial cut-off in

galactic disks, Astron. Astrophys. 407 (2003) L59–L62. doi:10.1051/
0004-6361:20031052. arXiv:astro-ph/0307389.

[206] R. P. Olling, The Highly Flattened Dark Matter Halo of NGC

4244, The Astronomical Journal 112 (1996) 481. doi:10.1086/118029.
arXiv:astro-ph/9605111.

199
[207] J. F. Becquaert, F. Combes, The 3D geometry of Dark Matter Halos.,
Astron. Astrophys. 325 (1997) 41–56. doi:10.48550/arXiv.astro-ph/
9704088. arXiv:astro-ph/9704088.

[208] W. J. G. de Blok, S. S. McGaugh, V. C. Rubin, High-Resolution

Rotation Curves of Low Surface Brightness Galaxies. II. Mass Models,
The Astronomical Journal 122 (2001) 2396–2427. doi:10.1086/323450.

[209] Y. Sofue, Dark halos of M 31 and the Milky Way, Publ. of

the Astron. Soc. of Japan 67 (2015) 75. doi:10.1093/pasj/psv042.
arXiv:1504.05368.

[210] P. Li, F. Lelli, S. McGaugh, J. Schombert, A Comprehensive Catalog

of Dark Matter Halo Models for SPARC Galaxies, The Astrophysical
Journal Supplement 247 (2020) 31. doi:10.3847/1538-4365/ab700e.
arXiv:2001.10538.

[211] R. Jimenez, L. Verde, S. P. Oh, Dark halo properties from rotation

curves, Mon. Not. R. Astron. Soc. 339 (2003) 243–259. doi:10.1046/
j.1365-8711.2003.06165.x. arXiv:astro-ph/0201352.

[212] A. Burkert, The Structure of Dark Matter Halos in Dwarf Galax-

ies, Astrophys. J. Lett. 447 (1995) L25–L28. doi:10.1086/309560.
arXiv:astro-ph/9504041.

[213] J. Bailin, M. Steinmetz, Internal and External Alignment of the Shapes

and Angular Momenta of ΛCDM Halos, Astrophys. J. 627 (2005) 647–
665. doi:10.1086/430397. arXiv:astro-ph/0408163.

[214] P. Bett, V. Eke, C. S. Frenk, A. Jenkins, J. Helly, J. Navarro,

The spin and shape of dark matter haloes in the Millennium sim-
ulation of a Λ cold dark matter universe, Mon. Not. R. Astron.
Soc. 376 (2007) 215–232. doi:10.1111/j.1365-2966.2007.11432.x.
arXiv:astro-ph/0608607.

[215] C. A. Vera-Ciro, L. V. Sales, A. Helmi, C. S. Frenk, J. F. Navarro,

V. Springel, M. Vogelsberger, S. D. M. White, The shape of dark
matter haloes in the Aquarius simulations: evolution and memory,
Mon. Not. R. Astron. Soc. 416 (2011) 1377–1391. doi:10.1111/j.
1365-2966.2011.19134.x. arXiv:1104.1566.

200
[216] K. T. E. Chua, A. Pillepich, M. Vogelsberger, L. Hernquist, Shape of
dark matter haloes in the Illustris simulation: effects of baryons, Mon.
Not. R. Astron. Soc. 484 (2019) 476–493. doi:10.1093/mnras/sty3531.
arXiv:1809.07255.

[217] J. Bailin, D. Kawata, B. K. Gibson, M. Steinmetz, J. F. Navarro, C. B.

Brook, S. P. D. Gill, R. A. Ibata, A. Knebe, G. F. Lewis, T. Okamoto,
Internal Alignment of the Halos of Disk Galaxies in Cosmological Hy-
drodynamic Simulations, Astrophys. J. Lett. 627 (2005) L17–L20.
doi:10.1086/432157. arXiv:astro-ph/0505523.

[218] B. Allgood, R. A. Flores, J. R. Primack, A. V. Kravtsov, R. H. Wech-

sler, A. Faltenbacher, J. S. Bullock, The shape of dark matter haloes:
dependence on mass, redshift, radius and formation, Mon. Not. R.
Astron. Soc. 367 (2006) 1781–1796. doi:10.1111/j.1365-2966.2006.
10094.x. arXiv:astro-ph/0508497.

[219] J. Prada, J. E. Forero-Romero, R. J. J. Grand, R. Pakmor, V. Springel,

Dark matter halo shapes in the Auriga simulations, Mon. Not.
R. Astron. Soc. 490 (2019) 4877–4888. doi:10.1093/mnras/stz2873.
arXiv:1910.04045.

[220] C. G. Palau, J. Miralda-Escudé, The oblateness of the Milky Way dark

matter halo from the stellar streams of NGC 3201, M68, and Palomar
5, Mon. Not. R. Astron. Soc. 524 (2023) 2124–2147. doi:10.1093/
mnras/stad1930. arXiv:2212.03587.

[221] J. Dubinski, D. Chakrabarty, Warps and Bars from the External Tidal
Torques of Tumbling Dark Halos, Astrophys. J. 703 (2009) 2068–2081.
doi:10.1088/0004-637X/703/2/2068. arXiv:0908.0168.

[222] D. R. Law, S. R. Majewski, K. V. Johnston, Evidence for a Triaxial

Milky Way Dark Matter Halo from the Sagittarius Stellar Tidal Stream,
Astrophys. J. Lett. 703 (2009) L67–L71. doi:10.1088/0004-637X/703/
1/L67. arXiv:0908.3187.

[223] H.-W. Rix, D. Zaritsky, Nonaxisymmetric Structures in the Stellar

Disks of Galaxies, Astrophys. J. 447 (1995) 82. doi:10.1086/175858.
arXiv:astro-ph/9505111.

201
[224] F. Bournaud, F. Combes, C. J. Jog, I. Puerari, Lopsided spiral galaxies:
evidence for gas accretion, Astron. Astrophys. 438 (2005) 507–520.
doi:10.1051/0004-6361:20052631. arXiv:astro-ph/0503314.
[225] D. Zaritsky, H. Salo, E. Laurikainen, D. Elmegreen, E. Athanassoula,
A. Bosma, S. Comerón, S. Erroz-Ferrer, B. Elmegreen, D. A. Gadotti,
A. Gil de Paz, J. L. Hinz, L. C. Ho, B. W. Holwerda, T. Kim, J. H.
Knapen, J. Laine, S. Laine, B. F. Madore, S. Meidt, K. Menendez-
Delmestre, T. Mizusawa, J. C. Muñoz-Mateos, M. W. Regan, M. Seib-
ert, K. Sheth, On the Origin of Lopsidedness in Galaxies as Deter-
mined from the Spitzer Survey of Stellar Structure in Galaxies (S4 G),
Astrophys. J. 772 (2013) 135. doi:10.1088/0004-637X/772/2/135.
arXiv:1305.2940.
[226] R. A. Angiras, C. J. Jog, A. Omar, K. S. Dwarakanath, Origin of disc
lopsidedness in the Eridanus group of galaxies, Mon. Not. R. Astron.
Soc. 369 (2006) 1849–1857. doi:10.1111/j.1365-2966.2006.10418.x.
arXiv:astro-ph/0604120.
[227] J. van Eymeren, E. Jütte, C. J. Jog, Y. Stein, R. J. Dettmar, Lop-
sidedness in WHISP galaxies. II. Morphological lopsidedness, As-
tron. Astrophys. 530 (2011) A30. doi:10.1051/0004-6361/201016178.
arXiv:1103.4929.
[228] R. H. M. Schoenmakers, M. Franx, P. T. de Zeeuw, Measuring non-
axisymmetry in spiral galaxies, Mon. Not. R. Astron. Soc. 292 (1997)
349–364. doi:10.1093/mnras/292.2.349. arXiv:astro-ph/9707332.
[229] J. van Eymeren, E. Jütte, C. J. Jog, Y. Stein, R. J. Dettmar, Lop-
sidedness in WHISP galaxies. I. Rotation curves and kinematic lopsid-
edness, Astron. Astrophys. 530 (2011) A29. doi:10.1051/0004-6361/
201016177. arXiv:1103.4928.
[230] P. M. W. Kalberla, L. Dedes, J. Kerp, U. Haud, Dark matter in
the Milky Way. II. The HI gas distribution as a tracer of the gravita-
tional potential, Astron. Astrophys. 469 (2007) 511–527. doi:10.1051/
0004-6361:20066362. arXiv:0704.3925.
[231] R. Rathulnath, C. J. Jog, Dynamics of prolate spheroidal mass dis-
tributions with varying eccentricity, Mon. Not. R. Astron. Soc. 433
(2013) 719–729. doi:10.1093/mnras/stt759. arXiv:1305.1408.

202
[232] K. Saha, E. S. Levine, C. J. Jog, L. Blitz, The Milky Way’s Dark
Matter Halo Appears to be Lopsided, Astrophys. J. 697 (2009) 2015–
2029. doi:10.1088/0004-637X/697/2/2015. arXiv:0903.3802.

[233] C. J. Jog, F. Combes, Lopsided spiral galaxies, Physics

Reports 471 (2009) 75–111. doi:10.1016/j.physrep.2008.12.002.
arXiv:0811.1101.

[234] M. D. Weinberg, Production of Milky Way Structure by the Magel-

lanic Clouds, Astrophys. J. Lett. 455 (1995) L31. doi:10.1086/309803.
arXiv:astro-ph/9507071.

[235] C. J. Jog, Dynamics of Orbits and Local Gas Stability in a Lopsided

Galaxy, Astrophys. J. 488 (1997) 642–651. doi:10.1086/304721.

[236] C. J. Jog, Dynamics in a Lopsided Spiral Galaxy: Negative Disk Re-

sponse to Perturbation Halo Potential, Astrophys. J. 522 (1999) 661–
668. doi:10.1086/307664.

[237] C. J. Jog, Large-scale asymmetry of rotation curves in lopsided spi-

ral galaxies, Astron. Astrophys. 391 (2002) 471–479. doi:10.1051/
0004-6361:20020832. arXiv:astro-ph/0207055.

[238] S. Ghosh, K. Saha, C. J. Jog, F. Combes, P. Di Matteo, Genesis of

morpho-kinematic lopsidedness in minor merger of galaxies, Mon. Not.
R. Astron. Soc. 511 (2022) 5878–5896. doi:10.1093/mnras/stac461.
arXiv:2105.05270.

[239] M. López-Corredoira, J. Betancort-Rijo, Azimuthal dependence of

the density distribution in outer galactic discs accreting intergalac-
tic flows, Astron. Astrophys. 493 (2009) L9–L12. doi:10.1051/
0004-6361:200811237. arXiv:0811.3875.

[240] A. Banerjee, C. J. Jog, The Flattened Dark Matter Halo of M31 as

Deduced from the Observed H I Scale Heights, Astrophys. J. 685 (2008)
254–260. doi:10.1086/591223. arXiv:0806.3610.

[241] R. Braun, The Distribution and Kinematics of Neutral Gas in M31,

Astrophys. J. 372 (1991) 54. doi:10.1086/169954.

203
[242] A. Banerjee, L. D. Matthews, C. J. Jog, Dark matter dominance at
all radii in the superthin galaxy UGC 7321, New Astronomy 15 (2010)
89–95. doi:10.1016/j.newast.2009.05.015. arXiv:0906.0217.

[243] W. J. G. de Blok, S. S. McGaugh, The dark and visible mat-

ter content of low surface brightness disc galaxies, Mon. Not.
R. Astron. Soc. 290 (1997) 533–552. doi:10.1093/mnras/290.3.533.
arXiv:astro-ph/9704274.

[244] A. V. Zasov, D. I. Makarov, E. A. Mikhailova, Thickness of Thin Stellar

Disks and the Mass of the Dark Halo, Soviet Astronomy Letters 17
(1991) 374.

[245] J. C. O’Brien, K. C. Freeman, P. C. van der Kruit, The dark

matter halo shape of edge-on disk galaxies. IV. UGC 7321, As-
tron. Astrophys. 515 (2010) A63. doi:10.1051/0004-6361/200912568.
arXiv:1002.3098.

[246] M. Das, R. Ianjamasimanana, S. S. McGaugh, J. Schombert, K. S.

Dwarakanath, Estimating the Oblateness of Dark Matter Halos Using
Neutral Hydrogen Velocity Dispersion, Astrophys. J. Lett. 946 (2023)
L8. doi:10.3847/2041-8213/acc10e. arXiv:2303.05104.

[247] E. M. Di Teodoro, F. Fraternali, 3D BAROLO: a new 3D algorithm

to derive rotation curves of galaxies, Mon. Not. R. Astron. Soc. 451
(2015) 3021–3033. doi:10.1093/mnras/stv1213. arXiv:1505.07834.

[248] F. Lelli, S. S. McGaugh, J. M. Schombert, M. S. Pawlowski, One

Law to Rule Them All: The Radial Acceleration Relation of Galax-
ies, Astrophys. J. 836 (2017) 152. doi:10.3847/1538-4357/836/2/152.
arXiv:1610.08981.

[249] B. S. Ryden, Oblate spheroidal mass distributions of varying ellipticity,

Mon. Not. R. Astron. Soc. 244 (1990) 341–348.

[250] K. Bekki, K. C. Freeman, Extended H I Spiral Structure and the

Figure Rotation of Triaxial Dark Halos, Astrophys. J. Lett. 574 (2002)
L21–L24. doi:10.1086/342262. arXiv:astro-ph/0206208.

204
[251] J.-M. Alimi, R. Koskas, The shape of dark matter halos: A new fun-
damental cosmological invariance, Astron. Astrophys. 691 (2024) A10.
doi:10.1051/0004-6361/202450845. arXiv:2406.15947.
[252] T. Veršič, M. Rejkuba, M. Arnaboldi, O. Gerhard, C. Pulsoni, L. M.
Valenzuela, J. Hartke, L. L. Watkins, G. van de Ven, S. Thater, Shapes
of dark matter haloes with discrete globular cluster dynamics: The
example of NGC 5128 (Centaurus A), Astron. Astrophys. 687 (2024)
A80. doi:10.1051/0004-6361/202349097. arXiv:2403.13109.
[253] C. J. Jog, Effective Q Criterion for Disk Stability in an External
Potential, The Astronomical Journal 147 (2014) 132. doi:10.1088/
0004-6256/147/6/132.
[254] S. Ghosh, C. J. Jog, Suppression of gravitational instabilities by
dominant dark matter halo in low-surface-brightness galaxies, Mon.
Not. R. Astron. Soc. 439 (2014) 929–935. doi:10.1093/mnras/stu013.
arXiv:1401.1271.
[255] V. P. Debattista, J. A. Sellwood, Warped Galaxies from Misaligned
Angular Momenta, Astrophys. J. Lett. 513 (1999) L107–L110. doi:10.
1086/311913. arXiv:astro-ph/9901153.
[256] J. J. Han, C. Conroy, L. Hernquist, A tilted dark halo origin of the
Galactic disk warp and flare, Nature Astronomy 7 (2023) 1481–1485.
doi:10.1038/s41550-023-02076-9. arXiv:2309.07209.
[257] M. Ideta, S. Hozumi, T. Tsuchiya, M. Takizawa, Time evolu-
tion of galactic warps in prolate haloes, Mon. Not. R. Astron.
Soc. 311 (2000) 733–740. doi:10.1046/j.1365-8711.2000.03092.x.
arXiv:astro-ph/9910030.
[258] P. Pranav, C. J. Jog, Response of a galactic disc to vertical perturba-
tions: strong dependence on density distribution, Mon. Not. R. Astron.
Soc. 406 (2010) 576–585. doi:10.1111/j.1365-2966.2010.16695.x.
arXiv:1003.3327.
[259] A. Kumar, M. Das, S. K. Kataria, The effect of dark matter
halo shape on bar buckling and boxy/peanut bulges, Mon. Not.
R. Astron. Soc. 509 (2022) 1262–1268. doi:10.1093/mnras/stab3019.
arXiv:2110.08165.

205
[260] R. Ibata, S. Chapman, A. M. N. Ferguson, G. Lewis, M. Irwin,
N. Tanvir, On the Accretion Origin of a Vast Extended Stellar Disk
around the Andromeda Galaxy, Astrophys. J. 634 (2005) 287–313.
doi:10.1086/491727. arXiv:astro-ph/0504164.
[261] A. Helmi, Is the dark halo of our Galaxy spherical?, Mon. Not.
R. Astron. Soc. 351 (2004) 643–648. doi:10.1111/j.1365-2966.2004.
07812.x. arXiv:astro-ph/0309579.
[262] C. Vera-Ciro, A. Helmi, Constraints on the Shape of the Milky Way
Dark Matter Halo from the Sagittarius Stream, Astrophys. J. Lett. 773
(2013) L4. doi:10.1088/2041-8205/773/1/L4. arXiv:1304.4646.
[263] R. Ibata, K. Malhan, W. Tenachi, A. Ardern-Arentsen, M. Bellazz-
ini, P. Bianchini, P. Bonifacio, E. Caffau, F. Diakogiannis, R. Errani,
B. Famaey, S. Ferrone, N. F. Martin, P. di Matteo, G. Monari, F. Re-
naud, E. Starkenburg, G. Thomas, A. Viswanathan, Z. Yuan, Charting
the Galactic Acceleration Field. II. A Global Mass Model of the Milky
Way from the STREAMFINDER Atlas of Stellar Streams Detected
in Gaia DR3, Astrophys. J. 967 (2024) 89. doi:10.3847/1538-4357/
ad382d. arXiv:2311.17202.
[264] A. Helmi, Velocity Trends in the Debris of Sagittarius and the Shape
of the Dark Matter Halo of Our Galaxy, Astrophys. J. Lett. 610 (2004)
L97–L100. doi:10.1086/423340. arXiv:astro-ph/0406396.
[265] A. Bonaca, A. M. Price-Whelan, Stellar streams in the Gaia era, New
Astronomy Review 100 (2025) 101713. doi:10.1016/j.newar.2024.
101713. arXiv:2405.19410.
[266] L. Posti, A. Helmi, Mass and shape of the Milky Way’s dark
matter halo with globular clusters from Gaia and Hubble, As-
tron. Astrophys. 621 (2019) A56. doi:10.1051/0004-6361/201833355.
arXiv:1805.01408.
[267] F. Combes, M. Arnaboldi, The dark halo of polar-ring galaxy NGC
4650a: flattened towards the polar ring?, Astron. Astrophys. 305 (1996)
763.
[268] S. A. Khoperskov, A. V. Moiseev, A. V. Khoperskov, A. S. Saburova,
To be or not to be oblate: the shape of the dark matter halo in polar

206
 ring galaxies, Mon. Not. R. Astron. Soc. 441 (2014) 2650–2662. doi:10.
1093/mnras/stu692. arXiv:1404.1247.
[269] M. Milgrom, A modification of the Newtonian dynamics as a possible
alternative to the hidden mass hypothesis., Astrophys. J. 270 (1983)
365–370. doi:10.1086/161130.
[270] M. Milgrom, MOND laws of galactic dynamics, Mon. Not.
R. Astron. Soc. 437 (2014) 2531–2541. doi:10.1093/mnras/stt2066.
arXiv:1212.2568.
[271] B. Famaey, S. S. McGaugh, Modified Newtonian Dynamics (MOND):
Observational Phenomenology and Relativistic Extensions, Liv-
ing Reviews in Relativity 15 (2012) 10. doi:10.12942/lrr-2012-10.
arXiv:1112.3960.
[272] I. Banik, H. Zhao, From Galactic Bars to the Hubble Tension: Weigh-
ing Up the Astrophysical Evidence for Milgromian Gravity, Symmetry
14 (2022) 1331. doi:10.3390/sym14071331. arXiv:2110.06936.
[273] F. J. Sánchez-Salcedo, K. Saha, C. A. Narayan, The thickness of HI
in galactic discs under MOdified Newtonian Dynamics: theory and
application to the Galaxy, Mon. Not. R. Astron. Soc. 385 (2008) 1585–
1596. doi:10.1111/j.1365-2966.2008.12941.x. arXiv:0712.0816.
[274] G. Gilmore, N. Reid, New light on faint stars - III. Galactic structure
towards the South Pole and the Galactic thick disc., Mon. Not. R.
Astron. Soc. 202 (1983) 1025–1047. doi:10.1093/mnras/202.4.1025.
[275] K. Vieira, G. Carraro, V. Korchagin, A. Lutsenko, T. M. Girard,
W. van Altena, Milky Way Thin and Thick Disk Kinematics with
Gaia EDR3 and RAVE DR5, Astrophys. J. 932 (2022) 28. doi:10.
3847/1538-4357/ac6b9b. arXiv:2205.00590.
[276] J. Y. Cheng, C. M. Rockosi, H. L. Morrison, Y. S. Lee, T. C.
Beers, D. Bizyaev, P. Harding, E. Malanushenko, V. Malanushenko,
D. Oravetz, K. Pan, K. J. Schlesinger, D. P. Schneider, A. Sim-
mons, B. A. Weaver, A Short Scale Length for the α-enhanced
Thick Disk of the Milky Way: Evidence from Low-latitude SEGUE
Data, Astrophys. J. 752 (2012) 51. doi:10.1088/0004-637X/752/1/51.
arXiv:1204.5179.

207
[277] J. Bovy, H.-W. Rix, C. Liu, D. W. Hogg, T. C. Beers, Y. S. Lee, The
Spatial Structure of Mono-abundance Sub-populations of the Milky
Way Disk, Astrophys. J. 753 (2012) 148. doi:10.1088/0004-637X/
753/2/148. arXiv:1111.1724.
[278] J. T. A. de Jong, B. Yanny, H.-W. Rix, A. E. Dolphin, N. F. Martin,
T. C. Beers, Mapping the Stellar Structure of the Milky Way Thick
Disk and Halo Using SEGUE Photometry, Astrophys. J. 714 (2010)
663–674. doi:10.1088/0004-637X/714/1/663. arXiv:0911.3900.
[279] M. Haywood, P. Di Matteo, M. D. Lehnert, D. Katz, A. Gómez, The
age structure of stellar populations in the solar vicinity. Clues of a two-
phase formation history of the Milky Way disk, Astron. Astrophys. 560
(2013) A109. doi:10.1051/0004-6361/201321397. arXiv:1305.4663.
[280] O. N. Snaith, M. Haywood, P. Di Matteo, M. D. Lehnert, F. Combes,
D. Katz, A. Gómez, The Dominant Epoch of Star Formation in the
Milky Way Formed the Thick Disk, Astrophys. J. Lett. 781 (2014) L31.
doi:10.1088/2041-8205/781/2/L31. arXiv:1401.1835.
[281] O. Snaith, M. Haywood, P. Di Matteo, M. D. Lehnert, F. Combes,
D. Katz, A. Gómez, Reconstructing the star formation history of the
Milky Way disc(s) from chemical abundances, Astron. Astrophys. 578
(2015) A87. doi:10.1051/0004-6361/201424281. arXiv:1410.3829.
[282] R. F. G. Wyse, G. Gilmore, Kinematics of the galaxy from a
magnitude-limited proper-motion sample., The Astronomical Journal
91 (1986) 855–869. doi:10.1086/114064.
[283] D. K. Ojha, O. Bienayme, A. C. Robin, V. Mohan, A multicolor survey
of absolute proper motions: galactic structure and kinematics in the
direction of the galactic center at intermediate latitude., Astron. Astro-
phys. 290 (1994) 771–784. doi:10.48550/arXiv.astro-ph/9407075.
arXiv:astro-ph/9407075.
[284] K. Fuhrmann, Nearby stars of the Galactic disc and halo - IV, Mon.
Not. R. Astron. Soc. 384 (2008) 173–224. doi:10.1111/j.1365-2966.
2007.12671.x.
[285] T. Bensby, S. Feltzing, M. S. Oey, Exploring the Milky Way stellar
disk. A detailed elemental abundance study of 714 F and G dwarf

208
 stars in the solar neighbourhood, Astron. Astrophys. 562 (2014) A71.
doi:10.1051/0004-6361/201322631. arXiv:1309.2631.
[286] J. Bovy, H.-W. Rix, D. W. Hogg, The Milky Way Has No Distinct
Thick Disk, Astrophys. J. 751 (2012) 131. doi:10.1088/0004-637X/
751/2/131. arXiv:1111.6585.
[287] S. Ghosh, F. Fragkoudi, P. Di Matteo, K. Saha, Bars and boxy/peanut
bulges in thin and thick discs. II. Can bars form in hot thick
discs?, Astron. Astrophys. 674 (2023) A128. doi:10.1051/0004-6361/
202245275. arXiv:2210.14244.
[288] J. J. Dalcanton, R. A. Bernstein, A Structural and Dynamical Study
of Late-Type, Edge-on Galaxies. II. Vertical Color Gradients and the
Detection of Ubiquitous Thick Disks, The Astronomical Journal 124
(2002) 1328–1359. doi:10.1086/342286. arXiv:astro-ph/0207221.
[289] F. Pinna, J. Falcón-Barroso, M. Martig, M. Sarzi, L. Coccato, E. Iodice,
E. M. Corsini, P. T. de Zeeuw, D. A. Gadotti, R. Leaman, M. Lyuben-
ova, R. M. McDermid, I. Minchev, L. Morelli, G. van de Ven, S. Vi-
aene, The Fornax 3D project: Unveiling the thick disk origin in FCC
170; possible signs of accretion, Astron. Astrophys. 623 (2019) A19.
doi:10.1051/0004-6361/201833193. arXiv:1901.04310.
[290] S. Comerón, H. Salo, J. H. Knapen, R. F. Peletier, The kine-
matics of local thick discs do not support an accretion origin, As-
tron. Astrophys. 623 (2019) A89. doi:10.1051/0004-6361/201833653.
arXiv:1901.10294.
[291] N. Scott, J. van de Sande, S. Sharma, J. Bland-Hawthorn, K. Free-
man, O. Gerhard, M. R. Hayden, R. McDermid, Identification of an
[α/Fe]—Enhanced Thick Disk Component in an Edge-on Milky Way
Analog, Astrophys. J. Lett. 913 (2021) L11. doi:10.3847/2041-8213/
abfc57. arXiv:2105.10649.
[292] V. Tsikoudi, Photometry and structure of lenticular galaxies. I - NGC
3115, Astrophys. J. 234 (1979) 842–853. doi:10.1086/157565.
[293] P. Yoachim, J. J. Dalcanton, Lick Indices in the Thin and Thick Disks
of Edge-On Disk Galaxies, Astrophys. J. 683 (2008) 707–721. doi:10.
1086/590246. arXiv:0805.4197.

209
[294] L. M. Widrow, S. Gardner, B. Yanny, S. Dodelson, H.-Y. Chen, Galac-
toseismology: Discovery of Vertical Waves in the Galactic Disk, As-
trophys. J. Lett. 750 (2012) L41. doi:10.1088/2041-8205/750/2/L41.
arXiv:1203.6861.

[295] L. M. Widrow, J. Barber, M. H. Chequers, E. Cheng, Bending and

breathing modes of the Galactic disc, Mon. Not. R. Astron. Soc. 440
(2014) 1971–1981. doi:10.1093/mnras/stu396. arXiv:1404.4069.

[296] S. Ghosh, V. P. Debattista, T. Khachaturyants, Age dissection of the

vertical breathing motions in Gaia DR2: evidence for spiral driving,
Mon. Not. R. Astron. Soc. 511 (2022) 784–799. doi:10.1093/mnras/
stac137. arXiv:2009.02343.

[297] M. D. Weinberg, Vertical Oscillation of the Galactic Disk, Astrophys.

J. 373 (1991) 391. doi:10.1086/170059.

[298] J. A. Sellwood, R. W. Nelson, S. Tremaine, Resonant Thickening of

Disks by Small Satellite Galaxies, Astrophys. J. 506 (1998) 590–599.
doi:10.1086/306280. arXiv:astro-ph/9805145.

[299] Gaia Collaboration, D. Katz, T. Antoja, M. Romero-Gómez, R. Drim-

mel, C. Reylé, G. M. Seabroke, C. Soubiran, C. Babusiaux, P. Di Mat-
teo, F. Figueras, E. Poggio, A. C. Robin, D. W. Evans, A. G. A. Brown,
A. Vallenari, T. Prusti, J. H. J. de Bruijne, C. A. L. Bailer-Jones,
M. Biermann, L. Eyer, F. Jansen, C. Jordi, S. A. Klioner, U. Lam-
mers, L. Lindegren, X. Luri, F. Mignard, C. Panem, D. Pourbaix,
S. Randich, P. Sartoretti, H. I. Siddiqui, F. van Leeuwen, N. A. Wal-
ton, F. Arenou, U. Bastian, M. Cropper, M. G. Lattanzi, J. Bakker,
C. Cacciari, J. Casta n, L. Chaoul, N. Cheek, F. De Angeli, C. Fabri-
cius, R. Guerra, B. Holl, E. Masana, R. Messineo, N. Mowlavi, K. Nien-
artowicz, P. Panuzzo, J. Portell, M. Riello, P. Tanga, F. Thévenin,
G. Gracia-Abril, G. Comoretto, M. Garcia-Reinaldos, D. Teyssier,
M. Altmann, R. Andrae, M. Audard, I. Bellas-Velidis, K. Benson,
J. Berthier, R. Blomme, P. Burgess, G. Busso, B. Carry, A. Cellino,
G. Clementini, M. Clotet, O. Creevey, M. Davidson, J. De Ridder,
L. Delchambre, A. Dell’Oro, C. Ducourant, J. Fernández-Hernández,
M. Fouesneau, Y. Frémat, L. Galluccio, M. Garcı́a-Torres, J. González-
Núñez, J. J. González-Vidal, E. Gosset, L. P. Guy, J. L. Halbwachs,

210
N. C. Hambly, D. L. Harrison, J. Hernández, D. Hestroffer, S. T.
Hodgkin, A. Hutton, G. Jasniewicz, A. Jean-Antoine-Piccolo, S. Jor-
dan, A. J. Korn, A. Krone-Martins, A. C. Lanzafame, T. Lebzelter,
W. Löffler, M. Manteiga, P. M. Marrese, J. M. Martı́n-Fleitas, A. Moit-
inho, A. Mora, K. Muinonen, J. Osinde, E. Pancino, T. Pauwels,
J. M. Petit, A. Recio-Blanco, P. J. Richards, L. Rimoldini, L. M.
Sarro, C. Siopis, M. Smith, A. Sozzetti, M. Süveges, J. Torra, W. van
Reeven, U. Abbas, A. Abreu Aramburu, S. Accart, C. Aerts, G. Al-
tavilla, M. A. Álvarez, R. Alvarez, J. Alves, R. I. Anderson, A. H.
Andrei, E. Anglada Varela, E. Antiche, B. Arcay, T. L. Astraat-
madja, N. Bach, S. G. Baker, L. Balaguer-Núñez, P. Balm, C. Barache,
C. Barata, D. Barbato, F. Barblan, P. S. Barklem, D. Barrado, M. Bar-
ros, M. A. Barstow, L. Bartholomé Muñoz, J. L. Bassilana, U. Bec-
ciani, M. Bellazzini, A. Berihuete, S. Bertone, L. Bianchi, O. Bien-
aymé, S. Blanco-Cuaresma, T. Boch, C. Boeche, A. Bombrun, R. Bor-
rachero, D. Bossini, S. Bouquillon, G. Bourda, A. Bragaglia, L. Bra-
mante, M. A. Breddels, A. Bressan, N. Brouillet, T. Brüsemeister,
E. Brugaletta, B. Bucciarelli, A. Burlacu, D. Busonero, A. G. Butke-
vich, R. Buzzi, E. Caffau, R. Cancelliere, G. Cannizzaro, T. Cantat-
Gaudin, R. Carballo, T. Carlucci, J. M. Carrasco, L. Casamiquela,
M. Castellani, A. Castro-Ginard, P. Charlot, L. Chemin, A. Chiavassa,
G. Cocozza, G. Costigan, S. Cowell, F. Crifo, M. Crosta, C. Crow-
ley, J. Cuypers, C. Dafonte, Y. Damerdji, A. Dapergolas, P. David,
M. David, P. de Laverny, F. De Luise, R. De March, R. de Souza,
A. de Torres, J. Debosscher, E. del Pozo, M. Delbo, A. Delgado, H. E.
Delgado, S. Diakite, C. Diener, E. Distefano, C. Dolding, P. Drazinos,
J. Durán, B. Edvardsson, H. Enke, K. Eriksson, P. Esquej, G. Eynard
Bontemps, C. Fabre, M. Fabrizio, S. Faigler, A. J. Falc a, M. Farràs
Casas, L. Federici, G. Fedorets, P. Fernique, F. Filippi, K. Findeisen,
A. Fonti, E. Fraile, M. Fraser, B. Frézouls, M. Gai, S. Galleti, D. Gara-
bato, F. Garcı́a-Sedano, A. Garofalo, N. Garralda, A. Gavel, P. Gavras,
J. Gerssen, R. Geyer, P. Giacobbe, G. Gilmore, S. Girona, G. Giuffrida,
F. Glass, M. Gomes, M. Granvik, A. Gueguen, A. Guerrier, J. Guiraud,
R. Gutié, R. Haigron, D. Hatzidimitriou, M. Hauser, M. Haywood,
U. Heiter, A. Helmi, J. Heu, T. Hilger, D. Hobbs, W. Hofmann,
G. Holland, H. E. Huckle, A. Hypki, V. Icardi, K. Janßen, G. Jevar-
dat de Fombelle, P. G. Jonker, Á. L. Juhász, F. Julbe, A. Karam-
pelas, A. Kewley, J. Klar, A. Kochoska, R. Kohley, K. Kolenberg,

211
 M. Kontizas, E. Kontizas, S. E. Koposov, G. Kordopatis, Z. Kostrzewa-
Rutkowska, P. Koubsky, S. Lambert, A. F. Lanza, Y. Lasne, J. B.
Lavigne, Y. Le Fustec, C. Le Poncin-Lafitte, Y. Lebreton, S. Leccia,
N. Leclerc, I. Lecoeur-Taibi, H. Lenhardt, F. Leroux, S. Liao, E. Li-
cata, H. E. P. Lindstrøm, T. A. Lister, E. Livanou, A. Lobel, M. López,
S. Managau, R. G. Mann, G. Mantelet, O. Marchal, J. M. Marchant,
M. Marconi, S. Marinoni, G. Marschalkó, D. J. Marshall, M. Martino,
G. Marton, N. Mary, D. Massari, G. Matijevič, T. Mazeh, P. J. McMil-
lan, S. Messina, D. Michalik, N. R. Millar, D. Molina, R. Molinaro,
L. Molnár, P. Montegriffo, R. Mor, R. Morbidelli, T. Morel, D. Mor-
ris, A. F. Mulone, T. Muraveva, I. Musella, G. Nelemans, L. Nicastro,
L. Noval, W. O’Mullane, C. Ordénovic, D. Ordóñez-Blanco, P. Os-
borne, C. Pagani, I. Pagano, F. Pailler, H. Palacin, L. Palaversa,
A. Panahi, M. Pawlak, A. M. Piersimoni, F. X. Pineau, E. Plachy,
G. Plum, E. Poujoulet, A. Prša, L. Pulone, E. Racero, S. Ragaini,
N. Rambaux, M. Ramos-Lerate, S. Regibo, F. Riclet, V. Ripepi,
A. Riva, A. Rivard, G. Rixon, T. Roegiers, M. Roelens, N. Rowell,
F. Royer, L. Ruiz-Dern, G. Sadowski, T. Sagristà Sellés, J. Sahlmann,
J. Salgado, E. Salguero, N. Sanna, T. Santana-Ros, M. Sarasso, H. Savi-
etto, M. Schultheis, E. Sciacca, M. Segol, J. C. Segovia, D. Ségransan,
I. C. Shih, L. Siltala, A. F. Silva, R. L. Smart, K. W. Smith,
E. Solano, F. Solitro, R. Sordo, S. Soria Nieto, J. Souchay, A. Spagna,
F. Spoto, U. Stampa, I. A. Steele, H. Steidelmüller, C. A. Stephen-
son, H. Stoev, F. F. Suess, J. Surdej, L. Szabados, E. Szegedi-Elek,
D. Tapiador, F. Taris, G. Tauran, M. B. Taylor, R. Teixeira, D. Terrett,
P. Teyssandier, W. Thuillot, A. Titarenko, F. Torra Clotet, C. Turon,
A. Ulla, E. Utrilla, S. Uzzi, M. Vaillant, G. Valentini, V. Valette, A. van
Elteren, E. Van Hemelryck, M. van Leeuwen, M. Vaschetto, A. Vecchi-
ato, J. Veljanoski, Y. Viala, D. Vicente, S. Vogt, C. von Essen, H. Voss,
V. Votruba, S. Voutsinas, G. Walmsley, M. Weiler, O. Wertz, T. Wev-
ers, L. Wyrzykowski, A. Yoldas, M. Žerjal, H. Ziaeepour, J. Zorec,
S. Zschocke, S. Zucker, C. Zurbach, T. Zwitter, Gaia Data Release
2. Mapping the Milky Way disc kinematics, Astron. Astrophys. 616
(2018) A11. doi:10.1051/0004-6361/201832865. arXiv:1804.09380.

[300] I. Carrillo, I. Minchev, G. Kordopatis, M. Steinmetz, J. Binney, F. An-

ders, O. Bienaymé, J. Bland-Hawthorn, B. Famaey, K. C. Freeman,
G. Gilmore, B. K. Gibson, E. K. Grebel, A. Helmi, A. Just, A. Kun-

212
 der, P. McMillan, G. Monari, U. Munari, J. Navarro, Q. A. Parker,
W. Reid, G. Seabroke, S. Sharma, A. Siebert, F. Watson, J. Wojno,
R. F. G. Wyse, T. Zwitter, Is the Milky Way still breathing? RAVE-
Gaia streaming motions, Mon. Not. R. Astron. Soc. 475 (2018) 2679–
2696. doi:10.1093/mnras/stx3342. arXiv:1710.03763.
[301] M. López-Corredoira, F. Garzón, H. F. Wang, F. Sylos Labini, R. Nagy,
Ž. Chrobáková, J. Chang, B. Villarroel, Gaia-DR2 extended kine-
matical maps. II. Dynamics in the Galactic disk explaining radial and
vertical velocities, Astron. Astrophys. 634 (2020) A66. doi:10.1051/
0004-6361/201936711. arXiv:2001.05455.
[302] H. F. Wang, M. López-Corredoira, Y. Huang, J. L. Carlin, B. Q. Chen,
C. Wang, J. Chang, H. W. Zhang, M. S. Xiang, H. B. Yuan, W. X. Sun,
X. Y. Li, Y. Yang, L. C. Deng, Mapping the Galactic disc with the
LAMOST and Gaia red clump sample: II. 3D asymmetrical kinematics
of mono-age populations in the disc between 6-14 kpc, Mon. Not.
R. Astron. Soc. 491 (2020) 2104–2118. doi:10.1093/mnras/stz3113.
arXiv:1905.11944.
[303] F. A. Gómez, I. Minchev, B. W. O’Shea, Y. S. Lee, T. C. Beers, D. An,
J. S. Bullock, C. W. Purcell, Á. Villalobos, Signatures of minor mergers
in the Milky Way disc - I. The SEGUE stellar sample, Mon. Not. R.
Astron. Soc. 423 (2012) 3727–3739. doi:10.1111/j.1365-2966.2012.
21176.x. arXiv:1201.5898.
[304] J. L. Carlin, J. DeLaunay, H. J. Newberg, L. Deng, D. Gole,
K. Grabowski, G. Jin, C. Liu, X. Liu, A. L. Luo, H. Yuan, H. Zhang,
G. Zhao, Y. Zhao, Substructure in Bulk Velocities of Milky Way Disk
Stars, Astrophys. J. Lett. 777 (2013) L5. doi:10.1088/2041-8205/
777/1/L5. arXiv:1309.6314.
[305] M. E. K. Williams, M. Steinmetz, J. Binney, A. Siebert, H. Enke,
B. Famaey, I. Minchev, R. S. de Jong, C. Boeche, K. C. Freeman,
O. Bienaymé, J. Bland-Hawthorn, B. K. Gibson, G. F. Gilmore, E. K.
Grebel, A. Helmi, G. Kordopatis, U. Munari, J. F. Navarro, Q. A.
Parker, W. Reid, G. M. Seabroke, S. Sharma, A. Siviero, F. G. Watson,
R. F. G. Wyse, T. Zwitter, The wobbly Galaxy: kinematics north and
south with RAVE red-clump giants, Mon. Not. R. Astron. Soc. 436
(2013) 101–121. doi:10.1093/mnras/stt1522. arXiv:1302.2468.

213
[306] A. Widmark, L. M. Widrow, A. Naik, Mapping Milky Way disk per-
turbations in stellar number density and vertical velocity using Gaia
DR3, Astron. Astrophys. 668 (2022) A95. doi:10.1051/0004-6361/
202244453. arXiv:2207.03492.

[307] R. Sancisi, Warped HI Disks in Galaxies, Astron. Astrophys. 53 (1976)

159.

[308] A. Bosma, 21-cm line studies of spiral galaxies. I. Observations of the

galaxies NGC 5033, 3198, 5055, 2841, and 7331., The Astronomical
Journal 86 (1981) 1791–1824. doi:10.1086/113062.

[309] V. Reshetnikov, F. Combes, Statistics of optical WARPS in spiral disks,

Astron. Astrophys. 337 (1998) 9–16. doi:10.48550/arXiv.astro-ph/
9806114. arXiv:astro-ph/9806114.

[310] C. A. Narayan, R.-J. Dettmar, K. Saha, Wobbly discs - corru-

gations seen in the dust lanes of edge-on galaxies, Mon. Not. R.
Astron. Soc. 495 (2020) 3705–3714. doi:10.1093/mnras/staa1400.
arXiv:2005.10290.

[311] D. Lynden-Bell, Free precession for the galaxy, Mon. Not. R. Astron.
Soc. 129 (1965) 299. doi:10.1093/mnras/129.3.299.

[312] C. Hunter, A. Toomre, Dynamics of the Bending of the Galaxy, As-

trophys. J. 155 (1969) 747. doi:10.1086/149908.

[313] M. D. Weinberg, Dynamics of an interacting luminous disc, dark halo

and satellite companion, Mon. Not. R. Astron. Soc. 299 (1998) 499–
514. doi:10.1046/j.1365-8711.1998.01790.x.

[314] M. D. Weinberg, L. Blitz, A Magellanic Origin for the Warp of the

Galaxy, Astrophys. J. Lett. 641 (2006) L33–L36. doi:10.1086/503607.
arXiv:astro-ph/0601694.

[315] F. A. Gómez, I. Minchev, B. W. O’Shea, T. C. Beers, J. S. Bullock,

C. W. Purcell, Vertical density waves in the Milky Way disc induced
by the Sagittarius dwarf galaxy, Mon. Not. R. Astron. Soc. 429 (2013)
159–164. doi:10.1093/mnras/sts327. arXiv:1207.3083.

214
[316] C. F. P. Laporte, I. Minchev, K. V. Johnston, F. A. Gómez, Foot-
prints of the Sagittarius dwarf galaxy in the Gaia data set, Mon. Not.
R. Astron. Soc. 485 (2019) 3134–3152. doi:10.1093/mnras/stz583.
arXiv:1808.00451.

[317] V. P. Reshetnikov, I. V. Chugunov, A. A. Marchuk, A. V. Mosenkov,

M. D. Kozlov, S. S. Savchenko, D. I. Makarov, A. V. Antipova, A. M.
Sypkova, Galactic warps: from cosmic noon to the current epoch,
arXiv e-prints (2025) arXiv:2504.12403. arXiv:2504.12403.

[318] S. Araki, A Theoretical Study of the Stability of Disk Galaxies and

Planetary Rings., Ph.D. thesis, Massachusetts Institute of Technology,
1985.

[319] Y. Revaz, D. Pfenniger, Bending instabilities at the origin of

persistent warps: A new constraint on dark matter halos, As-
tron. Astrophys. 425 (2004) 67–76. doi:10.1051/0004-6361:20041386.
arXiv:astro-ph/0406339.

[320] S. Khoperskov, P. Di Matteo, O. Gerhard, D. Katz, M. Haywood,

F. Combes, P. Berczik, A. Gomez, The echo of the bar buckling:
Phase-space spirals in Gaia Data Release 2, Astron. Astrophys. 622
(2019) L6. doi:10.1051/0004-6361/201834707. arXiv:1811.09205.

[321] F. H. Briggs, Rules of Behavior for Galactic WARPS, Astrophys. J.

352 (1990) 15. doi:10.1086/168512.

[322] G. Heald, G. Józsa, P. Serra, L. Zschaechner, R. Rand, F. Fraternali,

T. Oosterloo, R. Walterbos, E. Jütte, G. Gentile, The Westerbork
Hydrogen Accretion in LOcal GAlaxieS (HALOGAS) survey. I. Sur-
vey description and pilot observations, Astron. Astrophys. 526 (2011)
A118. doi:10.1051/0004-6361/201015938. arXiv:1012.0816.

[323] M. L. Sánchez-Saavedra, E. Battaner, E. Florido, Frequency of warped

spiral galaxies at visible wavelengths., Mon. Not. R. Astron. Soc. 246
(1990) 458.

[324] M. L. Sánchez-Saavedra, E. Battaner, A. Guijarro, M. López-

Corredoira, N. Castro-Rodrı́guez, A catalog of warps in spi-
ral and lenticular galaxies in the Southern hemisphere, Astron.

215
 Astrophys. 399 (2003) 457–467. doi:10.1051/0004-6361:20021751.
arXiv:astro-ph/0212026.
[325] H. B. Ann, J. C. Park, Warped disks in spiral galaxies, New Astronomy
11 (2006) 293–305. doi:10.1016/j.newast.2005.08.006.
[326] R. W. Nelson, S. Tremaine, The damping and excitation of galactic
warps by dynamical friction, Mon. Not. R. Astron. Soc. 275 (1995)
897–920. doi:10.1093/mnras/275.4.897. arXiv:astro-ph/9408068.
[327] M. H. Chequers, L. M. Widrow, Spontaneous generation of bending
waves in isolated Milky Way-like discs, Mon. Not. R. Astron. Soc. 472
(2017) 2751–2763. doi:10.1093/mnras/stx2165. arXiv:1708.06662.
[328] S. R. Kulkarni, C. Heiles, L. Blitz, Atomic hydrogen in the outer Milky
Way, Astrophys. J. Lett. 259 (1982) L63–L66. doi:10.1086/183848.
[329] J. Spicker, J. V. Feitzinger, Are there typical corrugation scales in our
Galaxy ?, Astron. Astrophys. 163 (1986) 43–55.
[330] E. Florido, E. Battaner, M. Prieto, E. Mediavilla, M. L. Sanchez-
Saavedra, Corrugations in the discs of spiral galaxies. NGC 4244 and
5023., Mon. Not. R. Astron. Soc. 251 (1991) 193–198. doi:10.1093/
mnras/251.2.193.
[331] E. Florido, E. Battaner, A. Gros, M. Prieto, E. Mediavilla, The optical
warp of NGC 5907, Astrophysics & Space Science 190 (1992) 293–301.
doi:10.1007/BF00644855.
[332] L. D. Matthews, J. M. Uson, Corrugations in the Disk of the Edge-
On Spiral Galaxy IC 2233, Astrophys. J. 688 (2008) 237–244. doi:10.
1086/592086. arXiv:0807.3560.
[333] L. D. Matthews, J. M. Uson, H I Imaging Observations of Superthin
Galaxies. II. IC 2233 and the Blue Compact Dwarf NGC 2537, The
Astronomical Journal 135 (2008) 291–318. doi:10.1088/0004-6256/
135/1/291. arXiv:0709.4249.
[334] T. Tepper-Garcı́a, J. Bland-Hawthorn, K. Freeman, Galactic seis-
mology: joint evolution of impact-triggered stellar and gaseous disc
corrugations, Mon. Not. R. Astron. Soc. 515 (2022) 5951–5968.
doi:10.1093/mnras/stac1926. arXiv:2204.12096.

216
[335] T. Asano, D. Kawata, M. S. Fujii, J. Baba, Growing local arm inferred
by the breathing motion, Mon. Not. R. Astron. Soc. 529 (2024) L7–L12.
doi:10.1093/mnrasl/slad190. arXiv:2310.02312.
[336] U. Banik, M. D. Weinberg, F. C. van den Bosch, A Comprehensive
Perturbative Formalism for Phase Mixing in Perturbed Disks. I. Phase
Spirals in an Infinite, Isothermal Slab, Astrophys. J. 935 (2022) 135.
doi:10.3847/1538-4357/ac7ff9. arXiv:2208.05038.
[337] V. P. Debattista, The vertical structure and kinematics of grand design
spirals., Mon. Not. R. Astron. Soc. 443 (2014) L1–L5. doi:10.1093/
mnrasl/slu069. arXiv:1405.6345.
[338] C. Faure, A. Siebert, B. Famaey, Radial and vertical flows induced
by galactic spiral arms: likely contributors to our ‘wobbly Galaxy’,
Mon. Not. R. Astron. Soc. 440 (2014) 2564–2575. doi:10.1093/mnras/
stu428. arXiv:1403.0587.
[339] G. Monari, B. Famaey, A. Siebert, Modelling the Galactic disc: per-
turbed distribution functions in the presence of spiral arms, Mon. Not.
R. Astron. Soc. 457 (2016) 2569–2582. doi:10.1093/mnras/stw171.
arXiv:1601.04714.
[340] G. Monari, B. Famaey, A. Siebert, The vertical effects of disc non-
axisymmetries from perturbation theory: the case of the Galactic bar,
Mon. Not. R. Astron. Soc. 452 (2015) 747–754. doi:10.1093/mnras/
stv1206. arXiv:1505.07456.
[341] A. Kumar, S. Ghosh, S. K. Kataria, M. Das, V. P. Debattista, Exci-
tation of vertical breathing motion in disc galaxies by tidally-induced
spirals in fly-by interactions, Mon. Not. R. Astron. Soc. 516 (2022)
1114–1126. doi:10.1093/mnras/stac2302. arXiv:2208.07096.
[342] T. Asano, M. S. Fujii, J. Baba, S. Portegies Zwart, J. Bédorf,
Ripples spreading across the Galactic disc: Interplay of direct
and indirect effects of the Sagittarius dwarf impact, arXiv
e-prints (2025) arXiv:2501.12436. doi:10.48550/arXiv.2501.12436.
arXiv:2501.12436.
[343] T. Antoja, A. Helmi, M. Romero-Gómez, D. Katz, C. Babusiaux,
R. Drimmel, D. W. Evans, F. Figueras, E. Poggio, C. Reylé, A. C.

217
 Robin, G. Seabroke, C. Soubiran, A dynamically young and per-
turbed Milky Way disk, Nature 561 (2018) 360–362. doi:10.1038/
s41586-018-0510-7. arXiv:1804.10196.

[344] L. M. Widrow, Swing amplification and the Gaia phase spirals, Mon.
Not. R. Astron. Soc. 522 (2023) 477–487. doi:10.1093/mnras/stad973.
arXiv:2302.14524.

[345] S. Tremaine, N. Frankel, J. Bovy, The origin and fate of the Gaia
phase-space snail, Mon. Not. R. Astron. Soc. 521 (2023) 114–123.
doi:10.1093/mnras/stad577. arXiv:2212.11990.

[346] J. Bland-Hawthorn, S. Sharma, T. Tepper-Garcia, J. Binney, K. C.

Freeman, M. R. Hayden, J. Kos, G. M. De Silva, S. Ellis, G. F. Lewis,
M. Asplund, S. Buder, A. R. Casey, V. D’Orazi, L. Duong, S. Khanna,
J. Lin, K. Lind, S. L. Martell, M. K. Ness, J. D. Simpson, D. B. Zucker,
T. Zwitter, P. R. Kafle, A. C. Quillen, Y.-S. Ting, R. F. G. Wyse, The
GALAH survey and Gaia DR2: dissecting the stellar disc’s phase space
by age, action, chemistry, and location, Mon. Not. R. Astron. Soc. 486
(2019) 1167–1191. doi:10.1093/mnras/stz217. arXiv:1809.02658.

[347] Z.-Y. Li, J. Shen, Dissecting the Phase Space Snail Shell, Astrophys. J.
890 (2020) 85. doi:10.3847/1538-4357/ab6b21. arXiv:1904.03314.

[348] T. Antoja, P. Ramos, F. López-Guitart, F. Anders, M. Bernet, C. F. P.

Laporte, Tidally induced spiral arm wraps encoded in phase space, As-
tron. Astrophys. 668 (2022) A61. doi:10.1051/0004-6361/202244064.
arXiv:2206.03495.

[349] T. Antoja, P. Ramos, B. Garcı́a-Conde, M. Bernet, C. F. P. Laporte,

D. Katz, The phase spiral in Gaia DR3, Astron. Astrophys. 673 (2023)
A115. doi:10.1051/0004-6361/202245518. arXiv:2212.11987.

[350] S. Tremaine, The geometry of phase mixing, Mon. Not. R. Astron.

Soc. 307 (1999) 877–883. doi:10.1046/j.1365-8711.1999.02690.x.
arXiv:astro-ph/9812146.

[351] N. Frankel, J. Bovy, S. Tremaine, D. W. Hogg, Vertical motion in

the Galactic disc: unwinding the snail, Mon. Not. R. Astron. Soc. 521
(2023) 5917–5926. doi:10.1093/mnras/stad908. arXiv:2212.11991.

218
[352] R. Guo, Z.-Y. Li, J. Shen, S. Mao, C. Liu, Measuring the Milky
Way Vertical Potential with the Phase Snail in a Model-independent
Way, Astrophys. J. 960 (2024) 133. doi:10.3847/1538-4357/ad037b.
arXiv:2310.10225.

[353] P. J. McMillan, J. Petersson, T. Tepper-Garcia, J. Bland-Hawthorn,

T. Antoja, L. Chemin, F. Figueras, S. Khanna, G. Kordopatis,
P. Ramos, M. Romero-Gómez, G. Seabroke, The disturbed outer
Milky Way disc, Mon. Not. R. Astron. Soc. 516 (2022) 4988–5002.
doi:10.1093/mnras/stac2571. arXiv:2206.04059.

[354] M. Bennett, J. Bovy, J. A. S. Hunt, Exploring the Sgr-Milky Way-disk

Interaction Using High-resolution N-body Simulations, Astrophys. J.
927 (2022) 131. doi:10.3847/1538-4357/ac5021. arXiv:2107.08055.

[355] W. H. Julian, A. Toomre, Non-Axisymmetric Responses of Dif-

ferentially Rotating Disks of Stars, Astrophys. J. 146 (1966) 810.
doi:10.1086/148957.

[356] P. Goldreich, D. Lynden-Bell, II. Spiral arms as sheared gravitational

instabilities, Mon. Not. R. Astron. Soc. 130 (1965) 125. doi:10.1093/
mnras/130.2.125.

[357] R. J. J. Grand, R. Pakmor, F. Fragkoudi, F. A. Gómez, W. Trick,

C. M. Simpson, F. van de Voort, R. Bieri, An ever-present Gaia snail
shell triggered by a dark matter wake, Mon. Not. R. Astron. Soc. 524
(2023) 801–816. doi:10.1093/mnras/stad1969. arXiv:2211.08437.

[358] H. Li, L. M. Widrow2024, Residuals of an equilibrium model for the

galaxy reveal a state of disequilibrium in the Solar Neighbourhood,
Mon. Not. R. Astron. Soc. 520 (2023) 3329–3344. doi:10.1093/mnras/
stad244. arXiv:2207.03516.

[359] I. Minchev, A. C. Quillen, M. Williams, K. C. Freeman, J. Nordhaus,

A. Siebert, O. Bienaymé, Is the Milky Way ringing? The hunt for
high-velocity streams, Mon. Not. R. Astron. Soc. 396 (2009) L56–L60.
doi:10.1111/j.1745-3933.2009.00661.x. arXiv:0902.1531.

[360] S. Khanna, S. Sharma, T. Tepper-Garcia, J. Bland-Hawthorn, M. Hay-

den, M. Asplund, S. Buder, B. Chen, G. M. De Silva, K. C. Freeman,

219
 J. Kos, G. F. Lewis, J. Lin, S. L. Martell, J. D. Simpson, T. Nord-
lander, D. Stello, Y.-S. Ting, D. B. Zucker, T. Zwitter, The GALAH
survey and Gaia DR2: Linking ridges, arches, and vertical waves in the
kinematics of the Milky Way, Mon. Not. R. Astron. Soc. 489 (2019)
4962–4979. doi:10.1093/mnras/stz2462. arXiv:1902.10113.

[361] J. A. S. Hunt, A. M. Price-Whelan, K. V. Johnston, R. L. McClure,

C. Filion, B. Cassese, D. Horta, Radial phase spirals in the Solar
neighbourhood, Mon. Not. R. Astron. Soc. 527 (2024) 11393–11403.
doi:10.1093/mnras/stad3918. arXiv:2401.08748.

[362] A. Widmark, C. Laporte, P. F. de Salas, Weighing the Galactic

disk using phase-space spirals. I. Tests on one-dimensional simula-
tions, Astron. Astrophys. 650 (2021) A124. doi:10.1051/0004-6361/
202140650. arXiv:2102.08955.

[363] S. Oh, M. Colless, F. D’Eugenio, S. M. Croom, L. Cortese, B. Groves,

L. J. Kewley, J. van de Sande, H. Zovaro, M. R. Varidel, S. Barsanti,
J. Bland-Hawthorn, S. Brough, J. J. Bryant, S. Casura, J. S. Lawrence,
N. P. F. Lorente, A. M. Medling, M. S. Owers, S. K. Yi, The SAMI
Galaxy Survey: the difference between ionized gas and stellar velocity
dispersions, Mon. Not. R. Astron. Soc. 512 (2022) 1765–1780. doi:10.
1093/mnras/stac509. arXiv:2202.10469.

[364] N. J. Adams, C. J. Conselice, L. Ferreira, D. Austin, J. A. A. Trus-

sler, I. Juodžbalis, S. M. Wilkins, J. Caruana, P. Dayal, A. Verma,
A. P. Vijayan, Discovery and properties of ultra-high redshift galax-
ies (9 ¡ z ¡ 12) in the JWST ERO SMACS 0723 Field, Mon. Not.
R. Astron. Soc. 518 (2023) 4755–4766. doi:10.1093/mnras/stac3347.
arXiv:2207.11217.

[365] S. Carniani, K. Hainline, F. D’Eugenio, D. J. Eisenstein, P. Jakob-

sen, J. Witstok, B. D. Johnson, J. Chevallard, R. Maiolino, J. M.
Helton, C. Willott, B. Robertson, S. Alberts, S. Arribas, W. M. Baker,
R. Bhatawdekar, K. Boyett, A. J. Bunker, A. J. Cameron, P. A. Cargile,
S. Charlot, M. Curti, E. Curtis-Lake, E. Egami, G. Giardino, K. Isaak,
Z. Ji, G. C. Jones, N. Kumari, M. V. Maseda, E. Parlanti, P. G. Pérez-
González, T. Rawle, G. Rieke, M. Rieke, B. R. Del Pino, A. Sax-
ena, J. Scholtz, R. Smit, F. Sun, S. Tacchella, H. Übler, G. Venturi,

220
 C. C. Williams, C. N. A. Willmer, Spectroscopic confirmation of two
luminous galaxies at a redshift of 14, Nature 633 (2024) 318–322.
doi:10.1038/s41586-024-07860-9. arXiv:2405.18485.

[366] M. Xiao, P. A. Oesch, D. Elbaz, L. Bing, E. J. Nelson, A. Weibel,

G. D. Illingworth, P. van Dokkum, R. P. Naidu, E. Daddi, R. J.
Bouwens, J. Matthee, S. Wuyts, J. Chisholm, G. Brammer, M. Dick-
inson, B. Magnelli, L. Leroy, D. Schaerer, T. Herard-Demanche,
S. Lim, L. Barrufet, R. Endsley, Y. Fudamoto, C. Gómez-Guijarro,
R. Gottumukkala, I. Labbé, D. Magee, D. Marchesini, M. Maseda,
Y. Qin, N. A. Reddy, A. Shapley, I. Shivaei, M. Shuntov, M. Ste-
fanon, K. E. Whitaker, J. S. B. Wyithe, Accelerated formation of
ultra-massive galaxies in the first billion years, Nature 635 (2024) 311–
315. doi:10.1038/s41586-024-08094-5. arXiv:2309.02492.

[367] J. Bland-Hawthorn, T. Tepper-Garcia, O. Agertz, C. Federrath, Turbu-

lent Gas-rich Disks at High Redshift: Bars and Bulges in a Radial Shear
Flow, Astrophys. J. 968 (2024) 86. doi:10.3847/1538-4357/ad4118.
arXiv:2402.06060.

[368] J. Lian, L. Luo, The Thickness of Galaxy Disks from z = 5 to 0

Probed by JWST, Astrophys. J. Lett. 960 (2024) L10. doi:10.3847/
2041-8213/ad1492. arXiv:2312.07070.

[369] S. S. Goyary, K. Saha, H. S. Singh, S. Sarkar, Bending instabilities of

m = 1 mode in disc galaxies: interplay between dark matter halo and
vertical pressure, Mon. Not. R. Astron. Soc. 526 (2023) 1138–1152.
doi:10.1093/mnras/stad2806. arXiv:2307.06063.

[370] M. Miyamoto, R. Nagai, Three-dimensional models for the distribution

of mass in galaxies., Publ. of the Astron. Soc. of Japan 27 (1975) 533–
543.

[371] W. Dehnen, S. Jafaritabar, Potential-density pairs for Galaxy discs

with exponential or sech 2 vertical profile, The Open Journal of Astro-
physics 8 (2025) 5. doi:10.33232/001c.128290. arXiv:2501.05916.

[372] L. Ferreira, C. J. Conselice, E. Sazonova, F. Ferrari, J. Caruana, C.-B.

Tohill, G. Lucatelli, N. Adams, D. Irodotou, M. A. Marshall, W. J.
Roper, C. C. Lovell, A. Verma, D. Austin, J. Trussler, S. M. Wilkins,

221
 The JWST Hubble Sequence: The Rest-frame Optical Evolution of
Galaxy Structure at 1.5 ¡ z ¡ 6.5, Astrophys. J. 955 (2023) 94. doi:10.
3847/1538-4357/acec76. arXiv:2210.01110.

[373] J. S. Kartaltepe, C. Rose, B. N. Vanderhoof, E. J. McGrath,

L. Costantin, I. G. Cox, L. Y. A. Yung, D. D. Kocevski, S. Wuyts,
H. C. Ferguson, M. B. Bagley, S. L. Finkelstein, R. O. Amorı́n, B. H.
Andrews, P. Arrabal Haro, B. E. Backhaus, P. Behroozi, L. Bisigello,
A. Calabrò, C. M. Casey, R. T. Coogan, M. C. Cooper, D. Croton,
A. de la Vega, M. Dickinson, A. Fontana, M. Franco, A. Grazian, N. A.
Grogin, N. P. Hathi, B. W. Holwerda, M. Huertas-Company, K. G.
Iyer, S. Jogee, I. Jung, L. J. Kewley, A. Kirkpatrick, A. M. Koeke-
moer, J. Liu, J. M. Lotz, R. A. Lucas, J. A. Newman, C. Pacifici,
V. Pandya, C. Papovich, L. Pentericci, P. G. Pérez-González, J. Pe-
tersen, N. Pirzkal, M. Rafelski, S. Ravindranath, R. C. Simons, G. F.
Snyder, R. S. Somerville, E. R. Stanway, A. N. Straughn, S. Tacchella,
J. R. Trump, J. Vega-Ferrero, S. M. Wilkins, G. Yang, J. A. Zavala,
CEERS Key Paper. III. The Diversity of Galaxy Structure and Mor-
phology at z = 3-9 with JWST, Astrophys. J. Lett. 946 (2023) L15.
doi:10.3847/2041-8213/acad01. arXiv:2210.14713.

[374] B. E. Robertson, S. Tacchella, B. D. Johnson, K. Hainline, L. Whitler,

D. J. Eisenstein, R. Endsley, M. Rieke, D. P. Stark, S. Alberts,
A. Dressler, E. Egami, R. Hausen, G. Rieke, I. Shivaei, C. C. Williams,
C. N. A. Willmer, S. Arribas, N. Bonaventura, A. Bunker, A. J.
Cameron, S. Carniani, S. Charlot, J. Chevallard, M. Curti, E. Curtis-
Lake, F. D’Eugenio, P. Jakobsen, T. J. Looser, N. Lützgendorf,
R. Maiolino, M. V. Maseda, T. Rawle, H. W. Rix, R. Smit, H. Übler,
C. Willott, J. Witstok, S. Baum, R. Bhatawdekar, K. Boyett, Z. Chen,
A. de Graaff, M. Florian, J. M. Helton, R. E. Hviding, Z. Ji, N. Ku-
mari, J. Lyu, E. Nelson, L. Sandles, A. Saxena, K. A. Suess, F. Sun,
M. Topping, I. E. B. Wallace, Identification and properties of intense
star-forming galaxies at redshifts z ¿ 10, Nature Astronomy 7 (2023)
611–621. doi:10.1038/s41550-023-01921-1. arXiv:2212.04480.

[375] V. Kuhn, Y. Guo, A. Martin, J. Bayless, E. Gates, A. Puleo, JWST

Reveals a Surprisingly High Fraction of Galaxies Being Spiral-like at 0.5
≤ z ≤ 4, Astrophys. J. Lett. 968 (2024) L15. doi:10.3847/2041-8213/
ad43eb. arXiv:2312.12389.

222
[376] H. Yan, B. Sun, C. Ling, An Edge-on Regular Disk Galaxy at z =
5.289, Astrophys. J. 975 (2024) 44. doi:10.3847/1538-4357/ad7de9.
arXiv:2407.04209.

[377] I. Trujillo, N. Chamba, J. H. Knapen, A physically motivated defi-

nition for the size of galaxies in an era of ultradeep imaging, Mon.
Not. R. Astron. Soc. 493 (2020) 87–105. doi:10.1093/mnras/staa236.
arXiv:2001.02689.

[378] P. G. van Dokkum, R. Abraham, A. Merritt, First Results from the

Dragonfly Telephoto Array: The Apparent Lack of a Stellar Halo in
the Massive Spiral Galaxy M101, Astrophys. J. Lett. 782 (2014) L24.
doi:10.1088/2041-8205/782/2/L24. arXiv:1401.5467.

[379] J. Zhang, R. Abraham, P. van Dokkum, A. Merritt, S. Janssens, The

Dragonfly Nearby Galaxies Survey. IV. A Giant Stellar Disk in NGC
2841, Astrophys. J. 855 (2018) 78. doi:10.3847/1538-4357/aaac81.
arXiv:1802.02583.

223