A Quick Look at the Diffuse Interstellar Medium

Abstract

To introduce our subject, we provide an overview of the interstellar medium (ISM) in our Galaxy and its relation to the stellar component. We then discuss the phase structure of the ISM, with a special emphasis on why the hydrogen takes the different forms it does. We then outline the observational signatures of the Hot Ionized, Warm Ionized, Warm Neutral, and Cold Neutral media, followed by a brief introduction to the molecular component of the Galaxy. A description of Photo-Dissociation Regions is presented with emphasis on the transition region from atomic to diffuse molecular gas. The role of dust and cosmic rays is also briefly described. Finally, we discuss the concept of a molecular cloud, a central theme of this book.

It is clear to everyone that astronomy at all events compels the soul to look upwards, and draws it from the things of this world to the other. — Plato

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1.1 Introduction

The interstellar medium or ISM is literally “the stuff between the stars”. Although stars dominate the baryonic component of a galaxy like the Milky Way, the gas and dust play a key role in their life cycle and in the evolution of galaxies. The ISM in spiral galaxies comprises primarily hydrogen and helium gas with a dust component that has less mass than the gas by a factor of 100, and a cosmic ray component that has an energy density similar to that of starlight.

In this book, we focus primarily on the lower density molecular component of the ISM, i.e., those regions where the volume density, n, is less than or about 10³ particles cm⁻³. In particular, we will examine the low-density, low-extinction molecular gas contained in the so-called diffuse and translucent molecular clouds and in the envelopes or interclump regions of Giant Molecular Clouds (GMCs). In this regime “molecular clouds” are often not very molecular, and so we will also discuss their atomic component. We will refer to these objects as the molecular part of the diffuse ISM and mostly ignore the ionized and very hot gas, even though these components are clearly diffuse. The molecular portion of the diffuse ISM does not play a direct role in the star formation process, but it provides the canvas against which the proto-star forming cores are projected or embedded. Moreover, it does play a key role in the formation of dark molecular clouds and GMCs, the sites of all star formation in the Galaxy.¹

To begin this exploration of the diffuse ISM, we will take a quick look at the ISM as a whole using the unifying concept of the Star-Gas Cycle. We then briefly discuss how the atomic gas component (including both neutral and ionized species) distinguishes itself by breaking itself up into distinct “phases” which have specific combinations of density and temperature. The basic reasons why this happens are discussed and then the observational signatures of the various phases are examined. We then take a quick look at the molecular component and try to put it in the context of the atomic phases by examining the role self-gravity and radiative self-shielding play in determining what form is taken by the molecular gas. Another unifying concept, the Photo-Dissociation Region or PDR (sometimes also known as a Photon-Dominated Region) is covered in some detail. The dust component is addressed next and then cosmic rays which, though very important in their own right, mostly play an indirect role in the diffuse ISM. We end this introductory chapter with a re-examination of the concept of a “molecular cloud”.

1.2 Overview of the ISM and Its Role in Spiral Galaxies

The gas in the ISM assumes several distinct, long-lived configurations of volume density ranging from 10⁻³ cm⁻³ to 10⁷ cm ⁻³ and temperature ranging from 10 K to 10⁷ K. With this wide range of temperatures and densities, the structures formed are very different in distribution, occupied galactic volume, morphology, relevant physical processes, and interplay with the stellar component. These density and temperature configurations or phases are descriptively called the Hot Ionized Medium (HIM), the Warm Ionized Medium (WIM), the Warm Neutral Medium (WNM), and the Cold Neutral Medium (CNM) (Kulkarni and Heiles 1987; Shull 1987). The latter component is the most dense with volume densities in the range 1 − 10² cm⁻³ and temperatures of order 10² K. The Cold Neutral Medium (CNM) is atomic, but an even denser and colder component of the ISM is mostly molecular and is considered separately because, unlike the atomic phases, self-gravity is a key feature in its dynamics. Dense concentrations of cold atomic and molecular gas are readily detected via centimeter- and millimeter-wave spectral techniques and are whimsically called clouds. It is from these predominantly molecular entities that stars form.

To determine the relative importance of the various gaseous components of the Galaxy, it is useful to inventory them in a typical location such as the Solar neighborhood (i.e., away from unique regions such as the Galactic Center or the outermost regions of the Galaxy). If the mass surface density is defined to be the total mass of material contained in a column whose footprint is a square parsec perpendicular to the Galactic plane, i.e., along the z-direction, then the stellar mass surface density, Σ _∗, of the Galaxy at the Solar Circle ( ∼ 8 kpc from the center) is about 50 M_⊙ pc⁻², Σ _HI ,² is 5 M_⊙ pc⁻², and $\varSigma _{H_{2}}$ is 1. 3 M_⊙ pc⁻² (Gould 1996; Liszt 1983; Dame et al. 1987). These values are typical of the solar vicinity; it is important to realize that the mass surface densities change significantly from the Galactic Center to the Outer Galaxy. For instance, in the inner 200 pc of the Galaxy, $\varSigma _{H_{2}}$ increases to several hundred M_⊙ pc⁻² (Morris and Serabyn 1996), while at distances greater than 18 kpc from the center $\varSigma _{H_{2}}$ is likely to be less than 0. 01 M_⊙ pc⁻² (Digel et al. 1994).

In spiral galaxies the ISM can make up to 20% of baryonic mass and there is a constant interplay between the stars and the gas and dust. The stellar component dominates the baryonic mass, contributing 10¹¹ M_⊙ in galaxies like the Milky Way. The mass in gas is lower by an order of magnitude from the stellar mass and the dust is down two orders of magnitude from the gas.

The interplay between the stars and the ISM is often referred to as the “star-gas cycle” or the baryonic cycle and is schematically depicted in Fig. 1.1. All stars, regardless of mass, form in molecular clouds. As the stars evolve they return gas to the ISM in the form of stellar winds. At the end of their lives, high-mass stars explode in supernova explosions, leaving behind supernova remnants. Low-mass stars return a significant amount of material in gentler fashion in the form of planetary nebulae. The dynamics of these mass exchanges are important because they involve supersonic processes which drive shocks and produce large cavities in the ISM. Eventually, the hot gas from the stellar winds, supernova remnants, and planetary nebulae cools, forming cold, neutral hydrogen clouds and a warmer, distributed, neutral hydrogen layer. A fraction of the CNM becomes molecular perhaps under the mediation of gravity, turbulence, and/or magnetic fields. Once molecular clouds form, the denser portions form stars and the process continues. Although the star-gas cycle is a superb organizing concept for understanding the interplay of stars and the ISM, it does not do justice to some important components of the ISM. Magnetic fields play an important role in structuring the ISM but are present only implicitly in the diagram. Similarly, cosmic rays, another important constituent of a galaxy, are accelerated by supernova explosions, but do not appear in the diagram. Finally, the role of the hot ionized gas, though present in the cycle, is probably not given its due as far as the global structure of a galaxy and its interactions with the exo-galactic environment are concerned. Despite these shortcomings, the star-gas cycle is critical to understanding the evolution of galaxies.

Fig. 1.1

The star-gas cycle in a galaxy. SNRs are supernova remnants; PNs are planetary nebula. These objects recycle stellar material into the ISM by way of their ejecta. “Winds” represent the recycled material produced by stellar winds during the normal lifetime of stars

1.2.1 A Remark on Physical Processes in the ISM

The galactic component that we will examine in this book is not as far from human experience as sometimes supposed. We actually come rather close to its densest parts when examining the ionosphere and exosphere of the Earth. Aside from the role of the planet’s gravity in maintaining mechanical balance on large scale, the densities (from a few to 10⁶ cm⁻³), the highly (but not completely) ionized state of the gas, and the presence of high energy particles in the form of precipitating electrons and protons from the magnetosphere, all resemble conditions we will encounter in the ISM. Further, the thermal state of the gas is governed by an external irradiating and ionizing source, at large distance (1 AU in this case is considerable compared to 1 R_⊕). The incident solar ultraviolet and X-ray photons are heating agents along with cosmic ray ionization, while recombination and collisional de-excitations cool the gas. It is optically thin to visible light, completely opaque to the Lyman continuum,³ and completely transparent to radio waves above a few tens of MHz.⁴ In short, this terrestrial gas is a close match to the diffuse ISM; they share many of the properties that make the ISM especially difficult to treat.

One of the most important distinctions between, say, a stellar atmosphere and these rarified environments is the role of collisions. When we use the word “temperature” to discuss the properties of the gas, note that it has many different meanings and should never be automatically taken to mean kinetic energy of a Maxwellian distribution of particles. We will return to this point in the following chapters.

1.3 How Does the ISM Manifest Itself in a Galaxy Like the Milky Way?

Even a quick look at the Milky Way on a dark, moonless night reveals dark patches in the stellar distribution such as the Aquila Rift or the Coalsack. In the nineteenth and early twentieth centuries there was spirited debate concerning why certain regions of the sky were devoid of stars. The two principal points of view consisted of postulating the existence of some obscuring medium between us and the background stars or that the dark patches were just “holes” in the heavens, that is, regions where the stellar density dropped precipitously.⁵ The answer turned out to be obscuration by dust and the ISM became an established part of the Galaxy. After the early work by Barnard and others, the availability of the plates of the Palomar Observatory Sky Survey (POSS) allowed Beverly Lynds (1962) to produce a catalog of so-called “dark clouds” which were later established to be molecular clouds. In the interim, the widespread distribution of cold atomic gas was established by the detection of the 21 cm spin-flip transition of neutral hydrogen—dubbed HI by astronomers (Ewen and Purcell 1951; Muller and Oort 1951). Absorption-emission observations (Radhakrishnan et al. 1972) implied that a portion of the 21 cm emission arose from a warmer, less dense, atomic component of the ISM and the theoretical idea of “phases”—regions of HI at different temperatures and densities, but roughly in pressure equilibrium—gained observational confirmation. The theory behind the phases is discussed in the next section but, besides dust and the gas, radio spectroscopic observations from the late 1960s and early 1970s showed a widespread molecular component along the Galactic plane. The key development in this area was the detection of the¹²C¹⁶O ground state rotational transition (J = 1 → 0) in 1970 at 115.271 GHz (Wilson et al. 1970). We will refer to this transition as just CO(1-0) throughout the rest of the book unless clarity dictates otherwise. Carbon monoxide is the second most abundant molecule in the ISM after H₂ and the CO(1-0) line almost immediately became the workhorse tracer of molecular gas in the Milky Way. Surveys of this transition along the Galactic plane established the broad properties of the molecular emission of the Galaxy (Stark 1979; Burton and Gordon 1978; Dame 1983; Sanders et al. 1984). By the early 1980s, ground based observations in the optical and radio regimes had established the existence of a pervasive ISM in the Milky Way consisting of well-mixed regions of gas and dust, with a widespread atomic component that was both warm and cold, and a denser molecular component which was localized in entities named “molecular clouds”. This latter component coincided with regions of star formation and the paradigm that stars came exclusively from collapsing clumps in molecular clouds was established.

The multiwavelength extension of the electromagnetic spectrum during the 1980s and 1990s and improved instrumentation enabled new forms of the ISM to be identified. Sensitive Hα observations with a Fabry-Perot spectrometer allowed the detection of a widespread warm hydrogen component that was ionized (Reynolds 1983). This differed from the well-known HII regions such as the Orion Nebula because the medium was at much lower densities and was distributed throughout the sky. Interestingly, its existence had long been suspected by pulsar astronomers who had attributed pulsar dispersion measures to a significant ionized component of the ISM.

A major breakthrough in directly imaging the dust component of the ISM came from the Infrared Astronomy Satellite (IRAS) launched in 1983 (Neugebauer et al. 1984). IRAS imaged the whole sky at 12, 25, 60, and 100 μm and revealed the thermal continuum emission from the dust component. Figure 1.2 shows the 100 μm emission over the whole sky. In addition to the extensive dust distribution along the Galactic plane, a wispy, low-level emission was detected at high Galactic latitudes ( | b | ≥ 20^∘) and dubbed the IRAS “cirrus” (Low et al. 1984).

Fig. 1.2

IRAS 100 μm emission from the whole sky at 2^′ resolution in an Aitoff projection in Galactic coordinates centered on the Galactic Center. The emission along the Galactic plane is primarily from thermal emission by dust particles in the ISM. The units are in MJy/ster and range from −1.68 x 10⁻⁸ to 1.43 x 10⁴ on a log scale. The black gaps in map represent regions where no data were taken. The image was made with the Skyview Virtual Observatory

It is not an exaggeration to say that the IRAS images revolutionized our understanding of the ISM because even low levels of dust were now visible and dust-gas comparisons could be made with much greater precision. While IRAS traced infrared emission in the mid- and far-infrared, the Two Micron All Sky Survey (2MASS) project imaged the sky in the near-infrared (1.24, 1.63, and 2.19 μm) from the ground and could trace dust absorbing background starlight (see Fig. 1.3).

Fig. 1.3

Three color (J, H, and K) 2MASS map of the whole sky in an Aitoff projection centered on the Galactic Center. The Galactic plane is clearly noticeable but, unlike Fig. 1.2, the infrared emission in the near infrared (1.24, 1.63. and 2.19 μm) is primarily from stellar photospheres so that the Galactic Bulge is evident. The dust component is present in this data in absorption as dark regions rather than emission as in the IRAS 100 μm data. The Large and Small Magellanic Clouds are prominent in the lower right-hand side of the image. Atlas image mosaic obtained as part of the Two Micron All Sky Survey (2MASS), a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by NASA and the NSF

Ultraviolet satellites (Copernicus, IUE, HST, FUSE, and GALEX) have revealed widespread spectral lines of OVI. X-ray satellites detected emission from more highly ionized species and the diffuse x-ray background, so the hottest component of the ISM could be traced. Although found primarily in the halo of the Galaxy, large cavities produced by supernovae seemed to permeate the ISM in the disk and the question of the morphology of the ISM vis-a-vis the cold, warm, and hot hydrogen components began to be debated with inconclusive results (see review by Cox 2005). The Galactic ISM can be seen even in gamma rays. Figure 1.4 shows an EGRET whole-sky image at energies > 100 MeV. The plane of the Galaxy is easily visible with the emission produced as cosmic rays collide with hydrogen nucleons and produce intermediate particles which decay and produce gamma-rays.

Fig. 1.4

EGRET gamma-ray emission at > 100 MeV from the whole sky in an Aitoff projection. The Galactic plane is clearly noticeable; this emission arises from cosmic rays colliding with hydrogen nucleons. Units are in counts/s/cm²/steradian and depicted on a log scale from 0 to 0.00422. The image was made with the Skyview Virtual Observatory

Although designed to study the Cosmic Microwave Background radiation (CMB), the WMAP satellite mapped the whole sky at 22, 30, 40, 60, and 90 GHz. Radio continuum emission from the Galaxy is clearly noticeable at all wavelengths (e.g., see Fig. 1.5) and has to be removed to allow study of the CMB.

Fig. 1.5

WMAP image of the whole sky in an Aitoff projection at 0. 88^∘ resolution at Q band (centered at 40 GHz—see Appendix B). Non-thermal emission from the Galactic plane is evident and the background fluctuations away from the plane are from the CBR. The units of the image are in milliKelvin and depicted on a log scale from −0.378 to 90.578. The image was made with the Skyview Virtual Observatory

More recently, the Spitzer Space Telescope and the Herschel Space Observatory have again revolutionized our view of the infrared sky. The Planck satellite, following on the heels of COBE and WMAP is giving new insights into the molecular distribution of the ISM in addition to its primary mission of studying the CMB.

With all this data available, the ISM can be studied in unprecedented detail across the electromagnetic spectrum. We have seen that there is a molecular, an atomic neutral, and an ionized component with the latter two taking many different temperature and density configurations known as phases. A basic question immediately arises: Why does the atomic and ionized component of the ISM break up into different phases? We turn to that question in the next section.

1.4 Why Does It Break Up into Phases?

As an initial approximation, the various distinct phases of the ISM are thought to be in thermal and pressure equilibrium and are produced when heating and cooling mechanisms in a region equilibrate at a particular temperature. A generalized energy loss function, $\mathcal{L}$ ( cgs units: erg s⁻¹ cm⁻³), is defined as the difference between the gas cooling function, Λ, and the gas heating rate, Γ. These functions depend on the composition and physical conditions of the medium (n and T). If the density dependence of the heating and cooling functions is included explicitly, $\mathcal{L}$ can be written as

$$\displaystyle{ \mathcal{L} = n^{2}\varLambda - n\varGamma \ \ \ \ . }$$

(1.1)

For values of the loss function greater than 0 the gas is cooling down, for values less than 0 the gas is heating up, and for $\mathcal{L} = 0$, the locus of the resulting points in the n-T plane defines thermal equilibrium.

In the low-density limit we can write the condition for thermal equilibrium as

$$\displaystyle{ \varGamma /(nkT) =\varLambda /(kT)\ \ \ \ . }$$

(1.2)

If heating is produced by cosmic rays and starlight, then at constant pressure Γ/P is constant and can have specific values of n and T for a given P. For typical values of Γ and P, there are four equilibrium points along the curve $\mathcal{L} = 0$ (see, e.g., Shull 1987, p. 225). Two of these are stable points and two are unstable (i.e., small perturbations in n and T will drive the gas parameters away from the original values). The criteria for determining stable vs. unstable equilibrium points is determined by examining stability criteria such as those derived by Field (1965) or Balbus (1986). A plot of Λ vs. T is known as the cooling curve (see, e.g., Sutherland and Dopita 1993) and the various stability criteria predict instability for temperature regimes where the cooling curve is steeply rising. The cooling curve as a function of T can be calculated if the abundance of the various elements in a region of the ISM is known.

The low-density ISM is heated primarily by stellar photons ejecting electrons from grains while the cooling occurs via radiation losses. For regions with densities around 1–10 cm⁻³ excitation of far-infrared fine-structure levels of ionized carbon (CII) dominates the cooling and produces a long-lived, stable phase with equilibrium temperatures ∼ 100 K, hence the name Cold Neutral Medium (CNM).⁶ The Warm Neutral Medium (WNM) exists at lower densities (10⁻¹ cm⁻³) and the cooling from neutral oxygen and the neutral hydrogen Lyman series produces theoretical equilibrium temperatures close to 10,000 K. However, Kulkarni and Heiles (1987) and Heiles and Troland (2003) find that about half of the WNM gas has temperatures of only a few thousand K. Gas at temperatures in this range is thermally unstable and so the simple physics described above does not seem to tell the whole story. Nevertheless, the assumption of stable phases in the ISM provides some insight into what is occurring there. The earliest models of the ISM involved the CNM and WNM and were known as “two-phase” models (Field et al. 1969). Cox and Smith (1974) proposed that supernova explosions radically change this picture because the remnants cool slowly, overlap, and produce a network of hot tunnels in the ISM. A few years later, McKee and Ostriker (1977) developed a supernova remnant-dominated model for the ISM consisting of three phases, the CNM, the WNM, and a hot (10⁶ K), low-density (10⁻³ cm⁻³) phase, the Hot Ionized Medium (HIM). Because of the long cooling times, the HIM dominates the volume of the galaxy in this model even though most of the gas mass is concentrated in the CNM. The HIM is therefore a background medium in which the other phases are embedded. A key component of the theory was the idea that mass is exchanged between the hot and cold phases of the ISM through the evaporation of cold gas when it comes into contact with the HIM, and the formation of dense, cold shells as supernova remnants expanded and cooled. McKee and Ostriker also predicted the existence of a low-density widespread ionized component which become known as the Warm Ionized Medium (WIM). When confronted with observations these theoretical ideas yielded some puzzling failures (see review by Cox 2005). Four decades after the seminal work by Field and Habing, the detailed nature of the ISM is still controversial as far as the morphology, mass, surface density with Galactic location, and interplay of the phases.

1.5 The Phases of the ISM: Observational Signatures

Given the vast range in temperature and density of the gas in the ISM it is not surprising that very different techniques are needed to probe the various phases. As a start, the temperature of the particular phase dictates in which region of the electromagnetic spectrum the gas is best observed. Thus, the hottest phase (the HIM) is detected in x-rays and by UV radiation. The coldest phase (the CNM and the molecular component) are best observed in the radio and infrared portions of the spectrum. The Warm Neutral Medium (WNM) presents some interesting problems as far as its direct detection, but the main probe is the HI 21 cm line. Finally, the Warm Ionized Medium (WIM) is observed by direct observations of the Hα line. We examine the techniques for observing and studying each of the phases in detail below.

1.5.1 The Hot Ionized Medium

Spitzer (1956) predicted the possibility that a hot phase of the ISM should be present surrounding the disk of the Galaxy. He reasoned that a hot, low-density ambient medium is required to maintain the neutral, high-velocity, HI clouds that had been detected far from the disk in what we now call the “halo” of the Galaxy. Once formed, this phase should be long-lived because its low density makes cooling difficult. It took nearly two decades before techniques became available to detect this gas. The advent of UV and x-ray detector technology on board satellites orbiting above the Earth’s atmosphere finally revealed the presence of a pervasive, low-density (n ∼ 10⁻³ cm⁻³), high-temperature (T ∼ 0.3–2 x 10⁶ K) medium surrounding the disk (Bowyer et al. 1968; Jenkins and Meloy 1974; York 1974). Gas at these temperatures is buoyant in the Galactic gravitational field and has a thermal scale height of several kpc. It is likely produced by supernova explosions and stellar winds from early-type stars heating and accelerating cold or warm HI by shocks. The gas is thought to exist in large, long-lived cavities (because of the relatively long cooling times) distributed throughout the disk (McKee and Ostriker 1977). Some of these cavities vent into the lower halo and the buoyant gas escapes the disk (see review by Cox 2005). Eventually, the gas cools and likely falls back onto the disk as in the “Galactic fountain” models of Shapiro and Field (1976) and Bregman (1980), but the constant activity within the disk ensures that the hot phase always blankets the disk. The exact morphology and distribution of the HIM in the disk is poorly known although its filling factor is thought to be as high as 0.5. The hot cavities and bubbles certainly exist (we live in one known as the Local Bubble), and many different cartoon models for the distribution of this gas have been proposed and are more or less plausible (see Savage 1987; Cox 2005; Shelton 2009). It is worth noting that additional sources of hot gas are required to account for the observed x-ray brightness, particularly at 0.7–0.8 keV. The currently considered source is infall from the intergalactic medium (Shelton 2006).

Ultraviolet satellites such as Copernicus, IUE, and FUSE provided the tools for direct detection of the HIM. Jenkins and Meloy (1974) and York (1974) used Copernicus to observe absorption lines of OVI at λ1032 and λ1038 Å along many lines of sight. In subsequent years, lines from other ions like NV, CIV, OVII, and OVIII established the existence of a hot diffuse component with temperature ranging from 10⁵ to 10⁷ K. The discovery of this component is what led McKee and Ostriker (1977) to propose their famous Three-Phase Model for the ISM. The development of x-ray satellites and detectors allowed the HIM to be routinely detected and studied from continuum and emission line observations. The resemblance of the physical conditions of the HIM to those of the outermost solar atmosphere lead to its alternate name, coronal halo gas.

1.5.2 The Warm Ionized Medium

Early-type (massive) stars power the HIM by mechanical injection of energy via stellar winds. In addition, their ionizing radiation produces another phase of the ISM: the Warm Ionized Medium or WIM. As the name implies, this phase has a significantly lower temperature ( ∼ 10⁴ K) and is produced as the Lyman continuum luminosity from these stars propagates throughout the disk and into the halo producing extensive regions of ionized hydrogen. The WIM differs from HII regions in that it is more extended and is comprised of lower density gas ( ∼ 0.1 cm⁻³ vs. 10² cm⁻³). Moreover, the SII/Hα ratio in the WIM is several times higher than in HII regions. This is characteristic of gas that is photoionized by a very weak radiation field from distant stars (Mathis 1986). Globally, 90% of the Galactic ionized hydrogen mass is contained in the WIM and not in HII regions and the overall mass in the WIM rivals that in molecular form (a few times 10⁹ M_⊙). Locally, the mass surface density is 2–3 M_⊙ pc⁻². Heiles et al. (1996) proposed that the WIM is composed of “chimneys” or “worms” (filamentary structures extending from the disk into the lower halo) that facilitate the transport of ionizing photons beyond the Galactic plane. As discussed above, the morphology of the ISM is not precisely known, but the large, low-density cavities somehow permit the ionizing radiation from these stars to reach regions that are hundreds of pc away from the stars, so that the scale height of the WIM is 1–2 kpc, less than the HIM, but significantly more than the neutral phases described below (see Reynolds and Ogden 1979; Norman and Ikeuchi 1989; Reynolds 1989; Gaensler et al. 2008). The WIM is also sometimes referred to as the Extended Low-Density Medium or ELDM (Mezger 1978).

The WIM was first postulated by Hoyle and Ellis (1963) as they analyzed the Galactic radio continuum spectrum from 1–100 GHz at high Galactic latitudes. In 1970, Gottesman and Gordon used radio recombination lines (RRLs) to detect an ionized medium not associated with HII regions. Pulsar dispersion measures (DM = ∫ n _e ds where n _e is the electron density and ds is the differential path length calibrated for distance) showed the existence of a distributed ionized component (Lyne et al. 1985; Lazio and Cordes 2002). If the distribution of pulsar dispersion measures and the ensuing estimates of n _e are assumed to have an exponential distribution, then the scale height of this component of the ISM is estimated to be ∼ 1 kpc. In addition, observations of the Hα recombination line can be used to estimate the emission measure (EM = ∫ n _e n _HII dV = ∫ n _e ² dV, where n _HII is the density of ionized gas) which depends on the temperature and density of ionized gas. The early radio work on emission measure also stemmed from observations of bremsstrahlung emission and absorption. More recently, RRLs have become powerful probes of the ionized gas in the ISM, including the WIM. RRLs do not suffer from interstellar extinction (see Sect. 1.8) and they are optically thin allowing for easier interpretation of the line profiles (see Sect. 2.1). Most importantly, they provide velocity information which can help locate the distance of the emitting region along the line of sight. RRLs are also excellent probes of the denser HII regions. In the infrared, fine structure lines of [NII] (at 122 and 205 μm) indicate that there is widespread ionized hydrogen, because nitrogen has an ionization potential of 14.532 eV and so where nitrogen is ionized, hydrogen must also be ionized. Other tracers of the WIM are the forbidden lines [SII] λ6716 and [NII] λ6584 (Madsen et al. 2006).

However, the most successful way to observe the WIM is to use high throughput Fabry-Perot spectroscopy to detect faint optical interstellar emission lines, particularly the Hα recombination line. This work was pioneered by Ron Reynolds at Wisconsin in the mid-1980s, so much so that the WIM is sometimes referred to as the “Reynolds Layer”. A dedicated Fabry-Perot on Kitt Peak, whimsically called WHAM (Wisconsin Hα Mapper), mapped the Hα emission throughout large regions of the Galaxy establishing temperatures (6000–10,000 K within 2–3 kpc of the midplane), densities (0.03–0.08 cm⁻³) and distribution of the phase (filling factor 0.2–0.4 within 2–3 kpc of the midplane - Reynolds 1991; Haffner et al. 2003; Hill et al. 2008). This phase of the ISM has also been detected in other galaxies, where it is more commonly known as the Diffuse Ionized Gas or DIG.

1.5.3 The Warm Neutral Medium

The neutral atomic gas is traced primarily by the 21 cm spin-flip transition of atomic hydrogen (1420.406 MHz). Line profiles of the 21 cm line show narrow features (FWHM < 10 km s⁻¹) thought to be fairly cold and dense ( ∼ 100 K; 10 cm⁻³) concentrated in filamentary or planar structures (discussed in the next sections), and broader features thought to represent a pervasive, warm (of order 10³ K), lower density ( ∼ 0.1 cm⁻³) phase known as the Warm Neutral Medium (WNM).⁷ Heiles and Troland (2003) derive a volume filling factor for the WNM of ≈ 0.5 with a minimum of 0.48 of the gas in the thermally unstable 500–5000 K region. Wolfire et al. (2003) find that the WNM has a thermal velocity dispersion of ∼ 8 km s⁻¹ implying a scale height of less than 200 pc.⁸ The heating of this medium is primarily photoelectric with the electrons ejected from dust grains or polycyclic aromatic hydrocarbons (PAHs), although cosmic rays also contribute directly to the heating. Cooling proceeds primarily by photon losses by visible transitions of oxygen and the Lyman-α line. The WNM was first detected observationally by emission/absorption observations of the 21 cm line (Radhakrishnan et al. 1972). By comparing single-dish or interferometric observations of HI absorption towards extragalactic sources with single-dish observations of HI emission along adjacent lines of sight, these authors noted that the absorption lines were narrow, while the adjacent emission lines had both narrow and broad components. The lack of broad absorption counterparts indicated that the broad emission had low opacity and, thus, was not very dense. Historically, the opacity was deemed to be inversely proportional to the excitation temperature of the 21 cm line (often referred to as the “spin temperature”—see, e.g., Kulkarni and Heiles 1987). Thus, the broad features are warm (and, conversely, the narrow features are relatively cold). In this manner, Radhakrishnan et al. (1972), Dickey et al. (1978) and Payne et al. (1983) conclusively showed the existence of the WNM as a pervasive phase of the ISM, likely the background “bath” in which the colder HI filaments and clouds are embedded. Although the conclusion is correct, the manner in which it was obtained is suspect as it now appears that the opacity-excitation temperature relationship (known as the τ _o − T _s relation) has little physical meaning (see Heiles and Troland 2003, for details).

The WNM can also be deduced directly from Gaussian decomposition of HI emission spectra. Mebold (1972) analyzed about 1300 HI emission spectra separating the profiles into narrow and wide Gaussian components. If the intrinsic velocity dispersion of the wide component (8.8 km s⁻¹) is interpreted as a kinetic temperature, an upper limit for the warm neutral medium of 9600 K is obtained. Other decomposition studies (e.g., Mebold et al. 1982; Verschuur and Schmeltz 1989; Verschuur and Magnani 1994; Alexander 2006; Haud and Kalberla 2007) broadly confirmed the original conclusions proposed by Mebold. Some of the later studies also found evidence of a very broad component (velocity dispersion > 25 km s⁻¹) whose width cannot be identified with a kinetic temperature and whose origin is unknown and may even be an instrumental artifact (e.g., see discussion of stray radiation; Sect. 4.2.5).

1.5.4 The Cold Neutral Medium

The 21 cm line emission that showed the presence of the WNM also provides clear evidence of a colder, denser phase that is not as broadly distributed. This is the CNM with temperatures of less than or around 100 K and densities of a few particles cm⁻³ up to tens of particles cm⁻³ (see Heiles and Troland 2003). Wolfire et al. (2003) find characteristic thermal velocities for this component ∼ 1.5 km s⁻¹. The main heating sources for this component are photoelectric heating from dust grains and cosmic rays (as for the WNM). Perhaps surprisingly, the primary coolant is the² P _3∕2 −² P _1∕2 transition of singly-ionized carbon at 158 μm. This is so for two main reasons: (1) The ionization potential is 11 eV. Therefore it is a substantial presence even if the hydrogen is predominantly neutral. (2) With h ν ∕k = 91 K, after the hyperfine states of hydrogen this fine-structure state of CI is the lowest-lying energy state of all gas-phase CNM constituents. Thus even very low energy electrons can collisionally excite this optically thin transition.

The CNM is distributed not so much in atomic “clouds” (though that was an early interpretation) as in filaments and sheets (Verschuur 1991; Heiles and Troland 2003). A significant fraction of this component is also associated with the envelopes of molecular clouds (e.g., Elmegreen and Elmegreen 1987). An excellent review of this phase of the ISM was given by Dickey and Lockman (1990), and we will not repeat their exposition. However, because molecular clouds are always associated with atomic gas, the relationship between the CNM and the molecular gas holds the key to understanding how the latter forms.

Although the CNM is the easiest phase of the ISM to observe (because of the intensity of the 21 cm line and the relatively low technology required to detect it), its morphology is still controversial, with various authors interpreting atomic clouds as filaments seen along their length while others claim that the filaments are sheets seen along the side. The association of a portion of this phase with the molecular gas makes it easier to discern its morphology as the molecular gas tends to be much more clumped and confined both spatially and in velocity. A clear problem in untangling the distribution of the CNM is angular resolution (see Fig. 1.6). Even with the 100-m Green Bank radiotelescope (GBT) and its 10^′ resolution at 21 cm, the relationship between the HI associated with the high-latitude cloud MBM 40 is not at all clear. At this resolution, the atomic hydrogen appears to be distributed in a cocoon enveloping the molecular core. However, with the factor of 3 greater resolution permitted by the 305-m Arecibo radiotelescope, the cocoon appears to be more like a partial atomic ring with molecular gas “filling in” the ring or perhaps forming a helix shape (Verschuur 1974; Shore et al. 2003). Given the difficulty in understanding the intricacies of the relationship between the cold atomic and molecular components, it is clear that observations at the highest spatial resolution are needed to understand the relationship between the two phases. Thus, the Square Kilometer Array (SKA) will likely be the key future telescope.

Fig. 1.6

The importance of angular resolution in observing the 21 cm HI lines. The top image is the 21 cm line integrated in velocity from 2–5 km s⁻¹ from a map of the high-latitude cloud, MBM 40, made with the 100 m GBT (10^′ resolution). The lower two figures are velocity slices 0.32 km s⁻¹ in width and centered at 2.6 and 3.2 km s⁻¹ made with the 305-m Arecibo radiotelescope (resolution 4^′)

At high Galactic latitudes, the optical depth of the 21 cm transition is small, so that the column density of HI is directly obtained from the observations. The relevant formula is

$$\displaystyle{ N(\mathrm{H}\mathrm{I}) = 1.823 \times 10^{18}\int T_{ B}\mathrm{dv}\ \ \ \ \ cm^{-2} }$$

(1.3)

where T _B is the brightness temperature (see section 4.2.6) of the 21 cm line and dv is the differential velocity over the line. However, the lines of sight through the Galactic disk can be several hundred parsecs, and untangling the radial distribution of the HI is often impossible. In particular, associating a fraction of the HI gas along a line of sight with a small molecular cloud is very difficult. Sometimes, a distinct HI spectral feature at the molecular cloud velocity is clearly seen [e.g., HI spectra from the Draco region taken by Goerigk et al. (1983) and Gir et al. (1994)], but, more often, there is no narrow spectral feature and the HI line extends over tens of km s⁻¹. Thus, deciding which fraction of the velocity range to associate with the molecular gas (typically with linewidths of less than a few km s⁻¹) is based only on plausibility arguments. Although some high-latitude clouds have been specifically identified by looking for isolated HI concentrations (e.g., Heiles et al. 1988), there is no obvious correlation between N(HI) and the presence of a high-latitude molecular cloud. In this respect, dust emission is a better signpost of the possible presence of molecular gas in a given region (see below).

1.6 Molecular Gas: Why Some of It Is a Phase and Some Isn’t

Thus far, we have been discussing atomic hydrogen, whether neutral or ionized. In the Galactic disk there exists a dense, clumped, molecular component which gives rise to stars and was historically traced primarily by extensive observations of the CO(1-0) line. We will discuss the global distribution and properties of this component in detail in Chap. 7 Here we are primarily concerned with how it relates to and differs from the atomic phases.

The molecular gas is found in the densest regions of the CNM and forms entities called “molecular clouds”. Although we do not want to discuss these objects in detail in this first chapter, for now we can think of them as coming in primarily two varieties: Giant molecular clouds with masses $\gtrsim \ 10^{4}$ M_⊙, and smaller molecular clouds. The latter—for the moment—are either dark clouds (interstellar clouds that contain significant obscuring dust and are primarily molecular in content), or diffuse clouds (where the dust extinction is low and atomic hydrogen is at least as abundant as the molecular gas). We will expand on these categories in Chap. 7

Unlike atomic structures, the majority of molecular clouds are gravitationally bound. Thus, Elmegreen (1993) proposed categorizing interstellar clouds of all types on the basis of their gravitational state: diffuse, self-gravitating, and unbound. In this schema, diffuse clouds are bound structures with insufficient mass to be held together by gravity. Instead, they are confined by external pressure. Their formation is likely due to shocks and thermal instabilities in which pressure is the important dynamical factor. In contrast, self-gravitating clouds contain enough mass to be gravitationally bound. Such clouds can form in regions of the interstellar medium where turbulence dissipates and dense gas becomes quiescent enough for gravity to be the most important structuring agent. Although these types of clouds are almost always molecular, there are some primarily atomic examples as well (see below). The third group, unbound clouds, are not confined by either pressure or gravity and are thus dispersing on about their sound-crossing time. They may form in the same manner as diffuse clouds, but with the high pressure dissipated. To distinguish between diffuse and self-gravitating clouds, Elmegreen introduced the dimensionless parameter P(Gσ _m ²)⁻¹ to effectively measure the importance of pressure versus gravity, Here, P is the external pressure, G is the gravitational constant, and σ _m is the average mass surface density of a cloud. If the parameter is significantly greater than 1, then gravity is relatively unimportant and the dynamics are dictated by external flows and pressure gradients. In contrast, if P(Gσ _m ²)⁻¹ ≪ 1, the cloud is self-gravitating and its internal dynamics are governed by internal pressures and density gradients. In the marginal case where P(Gσ _m ²)⁻¹ ∼ 1, the dynamical state of the cloud depends on which of the two terms begins to dominate. If the mass term dominates, the cloud tends to become self-gravitating on the free-fall timescale ( ∼ 10⁴ years). If P begins to dominate, the mass surface density quickly drops on the sound crossing timescale. Thus, the transition to one case or the other is expected to be relatively rapid compared to the lifetime of the cloud (10⁶–10⁷ years).

Interstellar clouds can be further divided into two sub-populations depending on their atomic/molecular fraction. For H₂ formation, dust is required to serve as a nucleation site for the molecule as gas phase reactions do not proceed rapidly enough under ISM conditions (see, e.g., Duley and Williams 1984). Once H₂ has formed, it must be shielded from the background diffuse UV radiation field, ϕ, as the molecule is readily photodissociated by the absorbed UV flux which is proportional to ϕ N ^1∕3. In terms of the volume density of hydrogen nucleons, n, the cloud size, D, the grain volume density, n _g , the grain cross-sectional area, σ _g , and the thermal velocity of the hydrogen, v _t , the hydrogen-grain collision rate is proportional to nn _g σ _g v _t . If we define the grain surface area per atom as $\mathcal{Z}$, then this rate is proportional to $n^{2}D\mathcal{Z}\sim nN\mathcal{Z}$. Molecular formation occurs when the path integral through the cloud of the hydrogen-grain collision rate is greater than the absorbed UV flux. The shielding function can then be written (following Elmegreen 1993) as

$$\displaystyle{ S = (n/60\ cm^{-3})(N/5 \times 10^{20}cm^{-2})^{2/3}(\mathcal{Z}/\mathcal{Z}_{ o})(\phi /\phi _{o})^{-1} }$$

(1.4)

where $\mathcal{Z}_{o}$ and ϕ _o are the values of the respective parameters in the solar neighborhood. Similarly, the dimensionless quantity, P(Gσ _m ²)⁻¹, can be parameterized as

$$\displaystyle{ P(G\sigma _{m}^{2})^{-1} = 15(n/60\ cm^{-3})(c_{ s}/1\ km\ s^{-1})^{2}(N/5 \times 10^{20}\ cm^{-2})^{-2} }$$

(1.5)

where c _s is the speed of sound in the cloud.

With these two quantity, clouds can be categorized depending on the relative values of S and P(Gσ _m ²)⁻¹. There are thus four cases: (1) If S is small and P(Gσ _m ²)⁻¹ is large, the objects with these characteristics have low column densities of atomic gas and so the cloud is predominantly atomic and diffuse as far as the volume density of gas is concerned. This situation represents a typical CNM entity. (2) If S is large and P(Gσ _m ²)⁻¹ is small, the column density of gas is now larger, so self-shielding allows the gas to transition from primarily atomic to molecular. Self-gravity of the cloud now dominates the dynamics and this object is a typical molecular cloud of the type seen in the large-scale CO surveys (e.g., Dame et al. 2001). Most CNM filaments or sheets and molecular clouds fall into these two categories. However, when n is small, but N is large, i.e., when a large mass of gas is spread out over a large volume, then we have case (3), both S and P(Gσ _m ²)⁻¹ are small. This would represent a self-gravitating cloud that is mostly atomic. An example of this type of structure could be the virialized, giant HI clouds observed in our Galaxy and in external galaxies (Elmegreen and Elmegreen 1987; Skillman 1987). Finally, case (4) is when both S and P(Gσ _m ²)⁻¹ are large so the objects are molecular, but self-gravity is not very important. Most of the diffuse molecular clouds which are the subject of this book fall in this category. Because this component is not self-gravitating, it has to interact with the atomic phases we have discussed and may constitute an unrecognized fifth phase of the ISM (because like the other four phases, gravity can be ignored); a cold, molecular medium (CMM).

1.7 The Transition from Atomic to Molecular Gas

Molecular clouds form from the neutral atomic hydrogen ISM. These have a wide range of size, densities, and molecular fraction that lead to very different types of structures which we will discuss in more detail in Chap. 7 Here some general considerations on the physics of the transition from atomic to molecular gas are presented that will be useful for better understanding the role these objects play in the ISM.

1.7.1 Photodissociation Regions: PDRs

Molecular formation in the diffuse ISM is the competition between gas-phase formation mechanisms and photodissociation. Because H₂ molecules self-shield by line-driven instead of continuum dissociation, they form a “sacrificial” stratum of molecules. A sufficient atomic gas column illuminated by UV radiation on only one side rapidly turns molecular once it is sufficiently optically thick in the photo-dissociating transition (some of the dissociating photons are also soaked up by the dust present in the region). This self-shielding and the ensuing precipitous rise in N(H₂) typically occurs at hydrogen column densities of ∼ 10¹⁹ cm⁻². Thus, the diffuse ISM can assume a layered structure; from the source of UV radiation inwards, a series of zones consisting of atomic gas, a transition region, and molecular gas quickly develops (on timescales of 10³ yr—see, e.g., Millar 2000). These layered regions are called Photo-dissociation Regions or PDRs (less commonly, a PDR is sometimes referred to as a Photon-Dominated Region) and their physical and chemical properties have been extensively discussed (Hollenbach and Tielens 1997, 1999, and references therein). When metals are included in the PDR models, a more complex layer structure develops with the outermost region (closest to the photodissociating source) consisting of HI and C⁺, an inner region where the carbon begins transitioning to C⁰, the beginnings of a molecular region with first-generation molecules such as OH and CH, but with the carbon primarily locked up in C⁰ and C⁺, a transition to a CO-dominated region, and, finally, a region where more complex molecules appear (see, e.g., Wolfire 2010). Note that the ionization potential of C⁰ is less than hydrogen.

The precise layer structure of PDRs requires numerical modeling but the basic parameter that governs the structure is the ratio of the photodissociation rate, which is proportional to G _o n(H _total ), where G _o is the standard intensity of the interstellar radiation field and n(H _total ) is the volume density of hydrogen nucleons (ionized, atomic, molecular), and the binary collisional rate for the creation of the various species in the PDR. These latter processes scale as n ² and so the parameter can be written G _o ∕n(H _total ). The diffuse ISM as a whole can be thought of as a PDR since it contains H, C⁰, and C⁺, and so can the outer regions of molecular clouds. Astronomers don’t normally think in this way because they usually study one particular layer of the PDR, ignoring the others. For example, the early studies of molecular clouds generally ignored the transition region to the atomic ISM, and focused almost exclusively on the densest regions where star formation was taking place.

The physical processes needed for calculations of PDR structure include the extinction of the interstellar radiation field (ISRF) by dust grains, and the microphysics of heating and cooling of the gas. The most important heating mechanisms in the various regions include photoelectric heating by grain ionization, H₂ ejection by grains, and H₂ excitation by UV photons. In the most opaque regions of the PDR (the dense molecular cores), the UV field is quenched and the heating is dominated by cosmic rays and x-rays. The cooling mechanisms include the radiative emission from the various species with CII dominating the cooling of the outer layers, giving way to CI, giving way to CO as the depth into the PDR increases.

Complicating matters further is the chemistry which depends sensitively on variations in the density and ISRF. We briefly discuss some aspects of astrochemistry relevant to the diffuse ISM in Chap. 3

1.7.2 Diffuse vs. Dark Clouds

Applying the PDR concept for molecular clouds, van Dishoeck and Black (1988) studied the photodissociation of CO in detail and arrived at a new way of categorizing molecular gas. Historically, the smaller molecular clouds had been divided into diffuse and dark molecular clouds on the basis of their opacity to background starlight. Before the 1990s, diffuse clouds had been the province of optical and near-UV astronomers, who detected them by observing narrow, optical, spectral lines seen against the background stellar continuum of an early-type star. It is worth remembering that the first detections of interstellar molecules were accomplished by this technique as CH, CH⁺, and CN transitions were detected in the late 1930s (Dunham and Adams 1937a,b; McKellar 1940). This technique precluded any mapping of the molecular structures around these special lines of sight so the discovery of actual molecular “clouds” (as opposed to molecule bearing lines of sight) had to wait until radio spectroscopy was developed after World War II. Dark clouds, cataloged by Barnard (1927) and, later, Lynds (1962) were known to contain molecules as early as 1963, when the OH lines were detected in emission by Weinreb et al. By 1970, the CO(1-0) line was detected (Wilson et al. 1970) and the picture of a dark cloud as a molecular cloud was already firmly established. Within a decade, the CO(1-0) transition had become the line of choice for detecting molecular gas in the Galaxy.

The relevant observational differences between diffuse and dark clouds included: the optical opacity, with A _V  < 1 for the diffuse clouds and greater than several for the dark clouds; N(H₂), with diffuse clouds prior to the 1990s believed to have N(H₂) < 10²⁰ cm⁻² and dark clouds > 10²¹ cm⁻²; and the CO/H₂ ratio, with diffuse clouds showing ratios < 10⁻⁶ and dark clouds ∼ 10⁻⁴. The bulk of the carbon in diffuse molecular clouds is in the form of C⁰ and C⁺, while in dark clouds the bulk is tied up in CO. In this early scheme, dictated primarily by the inability of millimeter-wave technology of the 1970s and early 1980s to detect molecular emission from the traditional diffuse clouds, an oversimplified idea was widespread: diffuse clouds are primarily atomic and dark clouds are primarily molecular. That this was insufficient should have been clear even then, since the work by Spitzer et al. (1973) and Savage et al. (1977) using the Copernicus UV data had established that the transition to a mostly molecular environment has a threshold of total hydrogen column density ≈ 5 x 10¹⁹ cm⁻² and E(B-V) ≈ 0.08 mag rather than 10²⁰ cm⁻² and 0.3 mag. These threshold objects were difficult to detect in CO until the mid-1980s so radio astronomers tended to ignore diffuse molecular clouds.⁹

As technology allowed the detection of CO emission from regions with A _V  ∼ 1 mag, Van Dishoeck and Black introduced a new category: the translucent cloud. These were differentiated by having visual extinctions between 1 and 5 magnitudes (in this scheme, diffuse clouds were defined to have A _V  < 1 mag and dark clouds > 5 mag) and column densities ∼ 10²⁰ cm⁻². More importantly, CO self-shields from photodissociation and is additionally shielded by H₂ and dust in the clouds. At values of N(H _total ) ≈ 1 × 10²⁰ cm⁻², the CO/H₂ abundance rises precipitously, from 10⁻⁶ to 10⁻⁴ even as the column density increases by less than an order of magnitude. In this sense, the translucent clouds are a transitional regime where N(CO) rises rapidly enough to allow detection of the lowest CO rotational transitions even with the technology of the 1980s. The exact column density range where the transition occurred depends on the radiation field, but, for most regions distant from O and B stars, this occurs at N(H₂) values of a few times 10²⁰ cm⁻². This scheme was certainly an improvement in categorizing small molecular clouds but still did not address the issue that significant, molecular-dominated entities could exist at A _V levels as low as 0.3 mag.

Most of the high-latitude clouds are thought to be translucent (but see remarks below) and it may be that a significant portion of the molecular gas in a galaxy (residing in the interclump medium of GMCs) is as well according to the definition of van Dishoek and Black (see Polk et al. 1988; Wright et al. 1992; Chiar et al. 1994). Moreover, some authors have dubbed all translucent gas a PDR because it represents the region where carbon transitions from primarily C⁰ and C⁺ to CO.

van Dishoeck and Black (1988) also considered the chemistry in diffuse gas and realized that in translucent regimes, the chemistry is also in transition from primarily photoprocesses (which dominate diffuse cloud chemistry), to collisional processes that dominate dark clouds. In translucent clouds, both types of processes are important.

This simple, and attractive, classification scheme for molecular clouds has however recently been challenged by Liszt et al. (2010). They find very strong CO lines in what can only be defined as diffuse molecular gas according to the definition of van Dishoeck and Black (1988). Perhaps the radiation field is weaker than normal in these objects so that the transition from C⁰ and C⁺ dominated to CO-dominated regions occurs at lower extinction, or perhaps other factors are involved. We will discuss this issue more deeply in Chaps. 7 and 8.

1.8 The Role of Dust

An excellent rule of thumb about the ISM is: Where there is gas, there is dust. Although its total Galactic mass is factor of 100 lower than the gaseous component, the dust is more than a trace constituent. With the advent of infrared astronomy, it is easier to detect than the gas and the resolution of the SDSS and 2MASS surveys, and observations from the Spitzer Space Telescope and the Herschel Space Observatory provide higher spatial resolution than spectral line maps from single-dish radio telescopes. In many respects, infrared observations of the dust constituent of molecular clouds have supplanted the spectral line maps of clouds so popular in the 1970s and 1980s.

The presence of dust in the ISM was first demonstrated by Trumpler (1930a,b,c). Even a perfunctory glance at a good photograph of the Milky Way will show regions with distinctly reduced numbers of stars (e.g., Barnard 1927). Nineteenth century astronomers dubbed these regions “dark nebulae” and Fig. 1.7 shows a prominent example. The pre-discovery can be assigned to John Herschel’s first glimpse of the Coalsack.

Fig. 1.7

The dark cloud Barnard 68. The quantity of dust in the cloud is great enough that it obscures all background stars. The proximity of the cloud to the Earth means that foreground stars are not present so that the region looks completely dark. The image was obtained with the 8.2-m VLT ANTU telescope and the multimode FORS1 instrument in March 1999. Credit: ESO

With the advent of infrared astronomy in the 1980s, emission from interstellar dust in the ISM was detected and studied (see Chap. 6). It soon became clear that the basic spectral properties of the dust on large scales could be explained by a large grain component (sizes ranging from 0.01 μm to 0.3 μm), a very small grain component (sizes less than 10 nm), and a component of very large molecules (PAHs). The large grains are in radiative equilibrium with the radiation field in the ISM and typically have temperatures of 17 K (a bit less in the denser, molecular regions of the ISM where the radiation field is attenuated). The other two components experience significant temperature fluctuations after photon absorption and emit over a wide range in the infrared below 60 μm.

Because the gas and dust are well-mixed, and since it is now relatively easy to image the entire sky in the infrared, dust emission is often used to trace the gas. Dust provides the nucleation site for molecular hydrogen formation (Duley and Williams 1984) so its role in producing the molecular component of the ISM must be addressed and we will do so in Chap. 6.

1.9 Cosmic Rays

Filling the diffuse medium, there is a population of superthermal charged particles whose energies require distinctive acceleration and propagation mechanisms, called cosmic rays (hereafter CR).¹⁰ The cosmic ray particles have several distinctive, universal properties. Their composition is mainly protons, electrons, and helium nuclei, all stripped bare. A heavier population of accelerated, completely charged nuclei reaches energies of tens to hundreds of MeV. In the standard parlance, primary cosmic rays are those originating at cosmic accelerators, for instance supernova explosions, where synchrotron emission by electrons evinces a process of energization above GeV (see Sect. 3.3.5 for further discussion). The mean CR energy density is inferred from synchrotron measurements although, strictly speaking, this applies only to the electrons. The sources are distributed throughout the Galaxy (Strong et al. 2007). These are mainly supernova remnants and pulsars, for the energy range below a few GeV, and confinement to the Galaxy results for energies less than TeV by diffusion of the particles in the turbulent magnetic field of the diffuse medium. Since this population is not thermalized, there is no reason to expect complete charge neutrality everywhere. Whatever its spatial distribution, the mean (or rms) magnetic energy density is assumed to be in rough equipartition with the relativistic particles. The synchrotron luminosity in some volume is the product of u _rel , the energy density in CR electrons, and B ², the magnetic energy density. Thus, for an observed nonthermal surface brightness, minimizing the sum of the two components with respect to the magnetic field, the equipartition argument gives a minimum magnetic pressure and, by extension, the CR density. This is about 1 eV cm⁻³ in the Galactic disk corresponding to a mean particle density of about 10⁻⁶ cm⁻³ at about 1 MeV, and similar values are obtained everywhere but in the Galactic center.

The energy spectrum of the photons is important as their number since the cross sections especially for ionization are strongly energy dependent. Below 10¹⁵ eV and above ∼ 1 GeV, the CR spectrum is a single power law, N(E) ∝ E ^−2. 7 for both electrons and protons. But at the lowest energies, below several hundred MeV, their energy is strongly modulated by the heliosphere. The Voyager interplanetary probes have now passed out of the heliosphere, beyond the immediate zone of modulation by the solar wind, and have measured the local CR energy distribution. For the N(E) ∝ E ^−α relation, they yielded an exponent of 1.45±0.15 below about 100 MeV and 3.15± 0.05 above that. The lower energy range is immediately relevant since this is the first direct determination of the energy spectrum in the interval that governs the chemistry. The contrast with near Earth and outer heliosphere measurements is striking. The flux for He nuclei is N(E) ≈ 1. 8(E∕GeV )^−2. 7 nucleons cm⁻² s⁻¹ GeV⁻¹ with approximately 80% of the primary CR nuclei being protons and about 1% electrons.

As we will discuss in later chapters, CR interactions with the diffuse gas are extremely important for governing the thermal and ionization state of the medium and its chemical makeup. They also provide a unique probe, through secondary γ-ray production, of the dust to gas ratio.

1.10 The Concept of a “Molecular Cloud”

GMCs were identified early during the Galactic plane surveys and were thought to have very well-defined edges (Blitz 1979). Thus, calling a localized concentration of molecular gas a “cloud” was a consequence of the observational methods. Average densities of GMCs are of order 10² cm⁻³ while the average density of the ISM is roughly 1 cm⁻³, so the idea of molecular clouds plowing through the ISM like cannonballs through air was quickly established. This mindset was reinforced by virial analyses of the stability of clouds which implied that clouds were long-lived, stable structures (see Sect. 4.3.1). Some estimates of the ages and lifetimes of clouds from the early 1980s were as high as > 10⁸ years (e.g., Scoville and Hersh 1979), and the idea of massive, long-lived entities that are distinct from their atomic surroundings took hold.

Of course, there were disturbing issues with this picture even then. GMCs were surrounded by large HI envelopes which had clearly played key roles in their formation (Elmegreen and Elmegreen 1987). Moreover, applying the PDR concept to these structures argued against a truly sharp boundary between molecular and non-molecular gas. An edge as traced by CO(1-0) does not require the absence of molecular hydrogen. The idea of gravitationally bound clouds also implied a too-high star formation rate for the Galaxy, so some type of supporting mechanism was needed to prevent the bulk of the gas in a GMC from collapsing and forming stars. It was thought a gravitational instability could be prevented by magnetic fields or turbulence. While a magnetically controlled cloud seemed dynamically reasonable, a turbulent molecular cloud posed a whole series of problems that “cannonballs” just do not address. Finally, a dissenting opinion on the lifetimes of GMCs was presented by Blitz and Shu (1980) who proposed 10⁷ years as a more reasonable age estimate.

Still unresolved is how low the molecular content of a cold hydrogen structure must be for the object to be called a molecular cloud. The cores of GMCs and dark clouds are easy; inside their HI envelopes the hydrogen nucleons are virtually all in the form of H₂. There is some cold HI that is well-mixed with the H₂ but the ratio of this atomic component to the molecular gas is ∼ 10⁻³-10⁻⁴ (Goldsmith and Li 2005). These objects are clearly molecular clouds in all senses. The situation is more complicated for less well-shielded regions with lower hydrogen column densities, such as the envelope or interclump medium of a GMC or in the diffuse and translucent clouds. It is here that the PDR models show that the HI content rises as one moves away from the opaque regions. With a visual extinction of 1 mag, it is very common to have sites where the H₂/H _total ratio is less than 0.5. Yet the objects show plenty of CO(1-0) emission and are considered “molecular” clouds even though there is more hydrogen nucleons in atomic form than locked in H₂. For the moment, we will call something a molecular cloud if it has a well defined boundary region where the CO(1-0) line can be detected in emission by a single-dish mm-wave radiotelescope with less than fifteen minutes of integration time given current (2017) mm-wave technology. For all practical purposes, these are regions with N(H₂) > several times 10¹⁹ cm⁻² and E(B-V) $\gtrsim$ 0.1 mag. Even excluding GMC cores and dark molecular clouds, this observational definition encompasses a vast zoo of objects; from those that are mostly atomic, diffuse and dominated by turbulence to those that are mostly molecular, gravitationally bound, and forming stars. Understanding the theoretical and observational differences between these groups is one of our principal themes.

Despite the caveats listed above, the idea of a molecular “cloud” as a distinct, stable, long-lived entity was the dominant paradigm throughout the 1980s and early 1990s. It wasn’t until numerical simulations studies of cloud structure revealed serious problem with molecular clouds supported by magnetic fields that the issue of what constitutes a mostly molecular interstellar entity was re-examined. We will discuss this development throughout the remainder of the book.