Interstellar V Molecules
Interstellar molecular gas was discovered in the 1940s with the observation of absorption bands from electronic transitions in CH, CH+, and CN superimposed on the spectra of bright stars. In the late 1960s, centimeter- and millimeter-wavelength radio observations detected emission from rotational transitions of OH (hydroxyl), CO (carbon monoxide), NH3 (ammonia), and H2CO (formaldehyde). The Copernicus satellite detected Far-UV absorption bands of H2 (the Lyman & Werner electronic transition bands) and HD in the early 1970s, and Lyman FUSE (launched in 1999) has detected diffuse H2 along nearly every line of sight it has looked (where it often gets in the way of lines from the extragalactic sources of interest to the observations!). Infrared observations beginning in the 1970s detected H2 in emission from forbidden rotational-vibrational transitions in the Near-IR, with advances in IR array detectors in the late 1990s greatly expanding studies of these transitions. It was in the 1980s, with the development of new millimeter receiver technology, that the study of interstellar molecules blossomed. Many molecular species were identified, and the field has grown sufficiently in depth that we can only give it the most basic treatment here. This section will focus on the basic physics of molecular line formation in the interstellar medium, and on the properties of giant molecular clouds. These notes assume a familiarity with basic molecular structure (electronic, vibrational, and rotational quantum states) as was covered in the Astronomy 823 course (Theoretical Spectroscopy). Persons using these notes outside of OSU can refer to standard texts (like Herzberg) for the necessary background information. While the dominant molecular species in the ISM is H2, because it is a homonuclear linear molecule with no permanent dipole moment all of the low-lying energy levels are quadrupole transitions with small transition probabilities (A-values) and relatively high excitation energies. The high excitation energies mean that these transitions are only excited at high temperatures or in strong UV radiation fields (i.e., fluorescence). Thus the most abundant molecule in the ISM, carrying most of the mass and playing a key role in excitation, thermal balance, and gas-phase chemistry, is virtually invisible to direct observation. As a consequence, most of what we know about interstellar molecules comes from observations of socalled tracer species, primarily CO which is observed in its J=10 rotational transition at =2.6 mm. This and other molecular species are observed as emission lines from pure rotational transitions at centimeter to millimeter wavelengths. Like we saw in the case with the HI 21-cm line, we must account for stimulated emission as well as collisional and radiative effects when deriving the line properties. Since the formation of molecular species like CO occurs under conditions favorable for H2 formation, we will try to estimate the amount of H2 from the observed amount of CO with the assistance of a few simplifying assumptions. A final aspect of the physics of molecular clouds is chemistry, both gas-phase and on the surfaces of dust grains. This is a rich topic that is sadly beyond the scope of this course.
V-1
Interstellar Molecules
Spectrum of Orion Molecular Cloud 1 (OMC1) in the 1.3mm band [Blake et al. 1987, ApJ, 315, 621] showing many of the 29 molecular species detected in this region.
V-1 Interstellar CO & Other Tracer Molecules
Radiative Transfer (yet again!)
In our discussion of the HI 21cm line our treatment of radiative transfer included the effects of stimulated emission. We were assisted in this analysis by the fact that the 21cm wavelength of the transition is in the long wavelength Rayleigh-Jeans limit of the background radiation field. For the rotational transitions of interstellar molecules like CO1 that occur at shorter wavelengths, this simplifying assumption is no longer true the analysis become more involved. In general, the background radiation field has a blackbody spectrum:
2h 3 1 I = B (TR ) = 2 h / kTR 1 c e
where TR is the brightness temperature of the background radiation field, which at millimeter wavelengths (e.g., for the CO J=10 transition at 2.6mm) is dominated by the cosmic microwave background radiation with TR2.725K. We know from COBE and WMAP observations that the cosmic microwave background is the most perfect blackbody yet observed.
I will follow the usual convention of denoting the principal isotopic species of atoms in a molecule without giving the atomic weights explicitly. For example, CO means 12C16O. The isotopic species will be given explicitly when comparing different isotopic forms (e.g., when discussing the 12CO/13CO ratio in the standard CO analysis). V-2
1
Interstellar Molecules
At millimeter wavelengths, the observed brightness temperature of a line Tb is often expressed in terms of an effective antenna temperature:
* TA = TA /
where is the receiver efficiency. A molecular line source is usually observed by chopping the telescopes beam between on- and off-source positions and measuring the difference in antenna temperatures. In general, the difference in brightness temperatures is
* TA = (1 e )
h 1 1 e h / kTexc 1 e h / kTR 1 k
The excitation temperature Texc for a given transition is defined as: nu gu h ul / kTexc =e nl gl For pure rotational transitions the excitation temperature is often called the rotation temperature, whereas for vibrational transitions it is called the vibration temperature. This nomenclature is analogous to the spin temperature defined for the HI 21cm hyperfine transition. The absorption optical depth is:
= ds = (nl Blu nu Bul ) I ds
On the right-hand side the first expression of the integrand is the pure absorption term, and the second is the stimulated emission term. Integrating the optical depth over the line profile and eliminating the Bs as we did previously in our treatment of atomic gas line transfer gives
h ul c2 ( Nl Blu Nu Bul ) = 2 Aul Nu eh / kTexc 1 8 ul c line where Nu is the column density of the molecules in the given upper excited state, and Aul is the transition probability for the line. There are three regimes of excitation temperature:
d =
(
)
* Texc = TR : TA = 0; no line is visible * Texc > TR : TA > 0; line appears in emission * Texc < TR : TA < 0; line appears in absorption
The excitation temperature will depend on the relative importance of radiative processes (which drive Texc towards TR) and collisional processes (which drive Texc towards the kinetic temperature of the gas). Collisions between the molecular species in question (CO etc.) and HI and/or H2 are the most important. Considering only a single collider for simplicity, the excitation temperature is: 1 Aul + Tkin nqul 1 = Texc Aul 1 + nq ul where: T* = h ul / k and nqul = collisional de-excitation rate TR 1 T* TR TR T*
V-3
Interstellar Molecules
Written this way, the excitation temperature is a harmonic mean between the radiation and kinetic temperatures weighted by the relative collisional and radiative de-excitation terms. The ratio of the radiative and collisional rates is just the ratio of the density to the critical density for the transition: A ncrit = ul qul as we have seen previously for UV, optical, and near-infrared forbidden lines. In the low-density limit (n < ncrit), the excitation temperature is driven towards the radiation temperature and no line will be visible. In the high-density limit (n>>ncrit), the excitation temperature is driven towards the kinetic temperature, and we see an emission line since Tkin > TR for most cases.
Critical Density and Line Visibility
At low densities, below the critical density, the excitation temperature will be only slightly above the radiation temperature and the emission line will be practically invisible. This is why you often read in the molecular line literature that a particular line is only visible at or above its critical density. At high densities, well above the critical density, the excitation temperature is at (or very near) the kinetic temperature of the gas, we often say that the line is thermalized. Because different molecular lines have different critical densities, line visibility can serve as an approximate density diagnostic. This rule of thumb is not without qualifications. Previously we noted that the weighting factor in the harmonic mean was the ratio of the density to the critical density. However, note that in the equation for 1/Texc, this factor is multiplied by an additional factor of (TR/T). If stimulated emission is important, T<<TR, and the density must be much larger than ncrit to produce visible emission lines. For example, the graph below shows the excitation temperature as a function of density for two molecular lines: the CS J=32 147 GHz line and the NH3 (J,K)=(1,1) 23 GHz line. This CS line has a critical density of 1.5106 cm3, and reaches an excitation temperature of 0.9Tkin at n=2106 cm3. However, the NH3 line has a critical density of 2103 cm3, but the excitation temperature does not approach the kinetic temperature until n>105 cm3, nearly 3 orders of magnitude larger. The reason is the greater importance of stimulated emission at low frequencies compared to at high frequencies. In general, if T=h/k<<Tkin, the density must be much larger than the critical density in order for the line to be visible. As such, the density must be very large compared to the critical density to thermalize lines at centimeter wavelengths (like the NH3 transition noted above), while at millimeter and submillimeter wavelengths, the lines of species like CO and CS are essentially thermalized at or near their critical densities. By the time you get to Infrared and Visible wavelengths stimulated emission becomes negligible and all collisionally excited lines thermalize at ncrit.
Texc vs. density for NH3 and CS transitions from Evans (1989)
V-4
Interstellar Molecules
Total Column Densities
Once the total optical depth and kinetic temperature have been measured, this only gives an estimate the column density for molecules in that particular rotational state. To convert this into a total column density, we can exploit the fact that in order for the line is visible at all it must be nearly thermalized. In the LTE approximation, the fractional population of a given rotational state, J, is given as:
g J e EJ / kT Q(T ) where Q(T) is the partition function, defined as the sum over rotational states. fJ =
Q(T ) = gi e Ei / kT
i
For rotational states in simple linear molecules like CO or CS, the statistical weights and energies of the Jth rotational state is: g J = (2 J + 1) EJ = J ( J + 1)hB (v) Here B(v) is a function of the moment of inertia associated with vibrational quantum number v of the molecule (rotational transitions come in ladders within a given vibrational state) B (v) = h 8 I (v)
2
where I(v) is the moment of inertia of the molecule when it is in vibrational quantum state v. The summation is relatively straightforward to perform as it can be truncated at a high J level because transition probabilities increase with increasing J, and transitions out of these higher states occur fast enough that they will be heavily depopulated and contribute negligibly to the partition function. At low density, the sum for Q(T) can be approximated as an integral:
Q(T ) (2 J + 1)e EJ / kT dJ As the system approaches LTE, the partition function is approximately Q (T ) kT hB (v) For example, for the CO J=10 transition, the LTE partition function is: T Q(T ) 2.76 K Strictly speaking, LTE is not always a valid assumption, and explicit non-LTE calculations are performed to solve the equations of statistical equilibrium to evaluate the partition functions. When you see a conversion between the observed CO J=10 flux to a total CO column density being computed in the LTE approximation, you will know that theyve followed the simple treatment described above.
Radiative Trapping
The preceding analysis has assumed that only photons from the background radiation field (TR) are used to compute the stimulated emission term, ignoring contributions from line photons. This assumption is only strictly true in an optically thin cloud. For strong lines in abundant molecular species, like the CO J=10 line, the cloud will be optically thick in these lines, and we will have to include stimulated emission from these emission-line photons in the radiative transfer treatment.
V-5
Interstellar Molecules
In an optically thick molecular cloud, line photons scatter around and are effectively "trapped". If the cloud is thick enough, the local density of line photons can greatly exceed the density of photons from the cosmic microwave background. These extra photons result in greater stimulated emission, driving the excitation temperature above the kinetic temperature of the gas. Further, the velocity field becomes important as we often observe that the widths of emission lines are larger than the expected thermal Doppler widths, either due to turbulent motions within the clouds or large-scale motions (e.g., velocity gradients or outflows). There are a number of ways to calculate the effects of radiative trapping, the most common is the escape probability treatment of Scoville & Solomon [1974, ApJ, 187, L67]. The details are beyond the scope of these notes, but the basic idea is that radiative trapping causes line photons to scatter many times about the cloud, enhancing the subsequent emergent line flux due to stimulated emission. Each emission-line photon emitted by normal downward radiative transitions can effectively multiply itself by stimulated emission before escaping the cloud. The result is that you can get a larger excitation temperature at a lower density than in an optically thin cloud, thus lowering the effective density at which the line appears to thermalize and become visible. Since CO J=10 line is always extremely optically thick in molecular clouds, some kind of radiative trapping treatment must be performed. For most Giant Molecular Clouds, the observed optical depths in the J=10 line are >10, lowering the effective density at which TexcTkin from 3000 cm3 to as low as 300 cm3.
Density & Temperature Diagnostics
Density Diagnostics
There are a variety of ways to estimate the density of a molecular cloud from observations of molecular lines, all of which come with various assumptions and caveats. Goldsmith gives a relatively complete list of the most common diagnostics in his review in Interstellar Processes along with citations to the key papers. The most basic way to estimate density is from the observed column density by assuming a path length through the cloud: n N tot L1 Ntot is the total column density along the line of sight, and L is the path length. A second way that is commonly encountered in the literature is to set lower limits on the density by noting which molecular lines of different critical densities are visible in a cloud. Some of the common emission lines (among many others) that are used in this way are: Transition
12
CO J=10
Frequency 115 GHz 23.7 GHz 49 GHz 4.3 GHz
ncrit(10K) ~1000 cm3 1800 cm3 4.6104 cm3 1.7105 cm3
NH3 (1,1) CS J=10 HCO+ J=10
If many lines from a given molecular species (e.g., H2CO) can be observed, a better method is to measure the ratios of lines that are not yet thermalized (n<ncrit). These line ratios will be density sensitive and may be used to estimate densities in roughly the same way that we used the [SII] and
V-6
Interstellar Molecules
[OII] doublet ratios at visible wavelengths in HII regions. Such ratios tend to have limited ranges of applicability, for example: Lines CO J=10 and J=23 CS J=76 and J=43 Applicable Range 10 < n(H2) < 104 cm3
2.5
105.5 < n(H2) < 107.5 cm3
The main limitation of using molecular line ratios as a density diagnostic is that detailed quantum mechanical calculations are required to work out the density dependence, and this has been done for only a few species with the sufficient precision. A final way of estimating densities that is sometimes used to assume that the cloud is in virial equilibrium, with turbulent motions holding the cloud up against self gravity. Interpreting the observed line width as the virial velocity, you estimate the virial mass and divide by the volume of the cloud to estimate a density. Crude, but it can give a robust mean density for the cloud.
Temperature Diagnostics
The most important thermometers in molecular clouds are the rotational transitions of 12CO. The low-J levels are very optically thick, and if the lines are thermalized, measuring the line strength gives the excitation temperature, and hence the kinetic temperature of the H2. In most molecular clouds without additional heating sources (i.e., away from regions of massive-star formation), the observed kinetic temperatures are in the range of 1020K. In these regions, the thermal balance is between heating by cosmic rays and collisions with warm (T=3040K) dust grains, and cooling by molecular line emission, with CO as one of the primary coolants. Near the edges of molecular clouds, heating by photoelectrons ejected by Interstellar Radiation Field photons becomes important. In clouds associated with massive-star formation (e.g., in Orion or M17), the cloud temperatures range from 4070K, and localized regions with heating sources can be as warm as a few 100K. For a few molecular line species, we can estimate the kinetic temperature directly, along with an estimate of the density, using line ratios of optically thin transitions. Emission lines of NH3, for example, are often used this way, in conjunction with detailed models. Again, the principal limitation of line ratio methods is the necessity of having using quantum mechanical models to predict the run of line ratios with density and temperature. Such data are known with sufficient precision for only a few molecular species.
Anomalous Excitation
Briefly, but not to diminish its importance, the complex quantum structure of molecules makes many of them susceptible to a variety of non-thermal excitation sources like radiative and collisional pumping. This can lead to anomalous under- or over-population of excited states if the combination of selection rules and transition probabilities works out just right. If a molecule can be pumped so that the population of an excited state greatly exceeds the population of the ground state (population inversion), the excitation temperature can greatly exceed the background radiation temperature, and the additional population of the excited state enhances the effect of stimulated emission. Under the right circumstances, this can lead to the production of an astrophysical maser or laser (in laboratory spectroscopy, you get a maser if the wavelength is longer than 1mm, and a laser if shorter than 1mm). The most common astrophysical masers are OH, H2O, and NH3 masers powered by either radiative or collisional pumping. These masers are usually
V-7
Interstellar Molecules
associated with regions of star formation, and in dusty outflows around AGB stars. This is an entirely separate field of endeavor beyond the scope of these notes, but the interested reader should consult Moshe Elitzurs review article in Interstellar Processes, or recent Annual Reviews articles on astrophysical masers. If the anomalous excitation leads to significant underpopulation of an upper level relative to the lower level of a molecular transition, the resulting excitation temperature can become lower than the background temperature, causing the line to appear in absorption as against the cosmic background. An example of this is the anomalous absorption seen in formaldehyde (H2CO), in which the upper level of the transition in question is depopulated by photon pumping into even higher levels. Collisional refrigeration mechanisms have also been proposed.
The Standard CO Analysis
As we saw before, the observed antenna temperature is (on-off) is: h * TA = (1 e ) [ f (Texc ) f (TR ) ] k where 1 f (T ) = h / kT e 1 Since the observed intensity depends on the excitation temperature, Texc, and the optical depth, , observations of a single line will not allow you to derive both [i.e., you have two unknowns but only one observable].
Because the energy of a rotational transition E(J) is inversely proportional to the moment of inertia of the molecule I(v), changes in the isotopic species of one (or both) atoms will change the line frequency sufficiently to distinguish the two isotopic forms. For the CO molecule, the isotopic ratio is 13 12 C/ C1/90, so that one expects the relative optical depth in the 13CO lines to be smaller than in the 12 CO lines. This expectation is at the heart of the standard CO analysis: in most cases the 12CO line is optically thick, while its isotopic partner 13CO is optically thin. Observations show that the typical relative brightnesses of the J=10 line is I(13CO)/I(12CO)~0.050.4, in agreement with this expectation. A further assumption is that 13CO and 12CO both arise in the same regions, and so share the same excitation temperature. If there are significant chemical fractionation effects, this assumption could be invalid (e.g., if local chemistry affects the creation/destruction of 13CO differently than that of 12CO). In the limit of large optical depth for the 12CO line, the excitation temperature is
h k h k ln 1 + * 12 TA ( CO) + TR Inserting the numbers for the J=10 line (h/k=5.53K): Texc =
Texc = 5.53 K 5.53 K ln 1 + * 12 TA ( CO) + TR
12
Here
* TA = observed effective line intensity of
CO
TR = background radiation field temperature (2.725K)
V-8
Interstellar Molecules
For 13CO, we assume that it is optically thin, and substitute in the Texc derived from the 12CO observations into the equation for the intensity of 13CO and solve for the optical depth in 13CO, (13CO). Because we expect that Texc>TR, the optical depth of the 13CO line is
T * ( 13 CO) ( CO) ln 1 A 12 * TA ( CO) Otherwise, we would have to solve the equations using the full form for the effective antenna temperature, which would need to be done numerically.
13
1
The optical depth in 13CO is then converted into a column density as follows: dv 1 e 5.53/Texc The integral is taken over the line profile expressed as a function of line-of-sight velocity v. This column density is often written as N(13CO), as one assumes that if the line is visible it must therefore be thermalized, and so TexcTkin and N is thus an LTE column density. N ( CO) = 2.85 10 cm
13 14 2
( 13 CO) [1 + Texc ]
N(12CO) is then derived by assuming an isotopic ratio for 13CO/12CO, usually the locally determined cosmic ratio of ~1/90.
V-9
Interstellar Molecules
V-2 Molecular Hydrogen (H2)
Molecular Hydrogen (H2) is the most abundant molecular species in the ISM, and plays a crucial role in cooling and molecular chemistry. Despite its importance, it is also the most difficult species to observe directly. The reason is that H2 is a simple, homonuclear molecule. Because it consists of two atoms of identical mass, the center of mass and the center of charge coincide, resulting in no permanent dipole moment. With no dipole moment, only quadrupole rotational transitions can occur. This means that only the J=0 and J=2 rotational transitions occur, while the J=1 (dipole) rotational transitions are strictly forbidden. This means that unlike other molecules we have studied, H2 emits no lines long-wavelength rotational lines. The small mass and small size of the H2 molecule gives it a low moment of inertia. This means that the first pure rotational transition, J=20, occurs at 28m, a part of the spectrum unobservable from the ground due to water-vapor absorption in the atmosphere. Further, h/k =514K for this transition, very large relative to typical temperatures in giant molecular clouds (1020K). The next pure rotational transition is J=42 at 12m, which has h/k 1200 K. The generally high energies of the first excited states of H2 means that we expect negligible H2 emission unless we are looking at unusually warm (500-1000K) H2 gas in proximity to hot stars or in regions of active star formation within or at the fringes of giant molecular clouds. Contrast this situation with the CO J=10 line at =2.6mm where h/k=5.53K and the molecule, which is easily excited by H2 or HI collisions at temperatures of T=1020K more typical of the cores of giant molecular clouds. In general, H2 is only directly observable as 1. Absorption at Far-UV wavelengths in the diffuse ISM along sight lines toward nearby stars in the Lyman and Werner band electronic transitions. These lines arise in both cold and warm H2. This is our only direct probe of the cold H2 gas that makes up most of the ISM. 2. Emission by Infrared rotational-vibrational transitions in the electronic ground state of H2 at wavelengths between 1 and 28m in relatively warm regions. The molecular gas must be warm (500-2000K), excited either by shocks, outflows, or UV fluorescence from nearby stars. We will examine each of these below.
UV Lyman & Werner Bands
H2 consists of 2 hydrogen atoms linked by a covalent bond (i.e., sharing electron pairs between the atoms). The lowest energy states of H2 are shown in the figure below from Field, Somerville, & Dressler [1966, ARAA, 4, 207]. Subject to the usual selection rules for permitted electronic transitions (=0,1; S=0) and symmetry constraints on H2, there are two sets of allowed transitions out of the H2 ground state:
Lyman Bands: E>11.2eV, <1108 (first band head)
+ X 1 + B 1 u g
Werner Bands: E>12.3eV, <1008 (first band head)
X 1 + C 1 u g
Excitation into higher electronic states of H2 requires photons with wavelengths shortward of 800. Since these photons are more likely to ionize HI (<912), these higher electronic transitions are generally suppressed. A complete list of transition probabilities and oscillator strengths for these
V-10
Interstellar Molecules
bands is given by Morton & Dinerstein 1976 [ApJ, 204, 1]. In all, the Lyman and Werner bands arising from the J=07 rotational levels in the ground vibrational state comprise nearly 400 absorption lines between 912 and 1120. An analogous set of HD (Deuterated Molecular Hydrogen) bands, shifted in wavelength, are also observed in this part of the spectrum. These, together with the HI and DI Lyman-series absorption lines, are potentially powerful diagnostics of atomic and molecular Hydrogen in the diffuse interstellar medium, a potential which until recently has gone unrealized.
Electronic states of H2, from Field, Somerville & Dressler (1966)
The electronic absorption bands of H2 occur in the Far-UV part of the spectrum inaccessible with either IUE or HST. Interstellar Lyman and Werner bands were first observed with the Copernicus satellite in 1974 (cf. Spitzer & Jenkins 1975, ARAA, 13, 133). Copernicus had very low sensitivity and could only detect these bands along relatively diffuse (e.g., AV<1) lines of sight (e.g., towards Puppis), although it did provide exceptional data. Later short-duration missions (IMAPS and ORFEUS) in the 1990s used small telescopes and their observations of the H2 bands were limited to bright stars with relatively low line-of-sight extinction, and only a few targets due to their short mission times. Sounding rockets were used to study specific targets for very brief times. This all changed with the launch of the FUSE satellite in June 1999. FUSE is ~105 times more sensitive than Copernicus in the Far-UV band, and has produced must of the best data on the H2 Lyman and Werner band absorption to date. This sensitivity has also allowed extension of these observations to so-called translucent line of sight, those with AV>1 mag (e.g., Snow et al. 2000, ApJ, 538, L65, Rachford et al. 2001, ApJ, 555, 839 & 2002, ApJ, 577, 221). This particular study was one of the key projects for the FUSE mission.
V-11
Interstellar Molecules
FUSE spectra of H2 Lyman/Werner absorption bands along a translucent line of sight [From Rachford et al. 2002]
These absorption lines are the only way at present to directly detect the cold H2 gas that makes up most of the ISM. The limitations of these bands as ISM diagnostics are the same as for the UV atomic absorption lines, namely that information is confined to particular lines of sight back lit by strong UV sources (e.g., OB stars or AGNs) with sufficiently low extinction that the UV light from the background source can be detected. However, because these absorption lines of H2 are so strong, the H2 Lyman/Werner bands have been seen along every line of sight where FUSE has observed, in fact they are often a foreground nuisance when observing extragalactic lines of sight, and in choosing targets one must take care not to pick sight lines where the strongest H2 lines are expected to be saturated, as the redshifted lines of interest in, for example, an AGN have been known to fall into these bands. The FUSE translucent cloud study produced direct measurements of the H2 column densities from the J=0 and J=1 transitions along lines of sight to more than 40 early-type stars with AV>1. Combined with earlier Copernicus data for diffuse lines of sight, the following general conclusions can be made: 1. The kinetic temperature of the H2 gas range between ~50100K, with a mean value around 70K. This temperature is uncorrelated with the amount extinction along a line of sight, but hotter H2 gas can be found among diffuse lines of sight (i.e., there is greater scatter in Tkin, extending to higher temperatures), with cooler temperatures prevailing in regions where the H2 column density becomes large enough for self shielding to become important. 2. The molecular fraction f H 2 = 2 N H 2 /(2 N H 2 + N HI ) is low (<104) for sight lines with
E(BV)<0.08, then it increases abruptly to >0.01 because of the onset of self-shielding, and rises to as high as 0.8 with considerable scatter in translucent sight lines. Nowhere, however, does the molecular fraction get as large as 90% as expected by earlier theoretical models. 3. While there is no detailed correlation between the molecular fraction and color excess or RV, the available data show that f H 2 is well correlated with the density-sensitive fractional CN abundance, the width of the 2175 bump, and the slope of the far-UV extinction, and a
V-12
Interstellar Molecules
general trend for f H 2 to be anticorrelated with the kinetic temperature (hotter H2 gas has a smaller molecular fraction). The relations with temperature and density are unsurprising: colder, denser clouds are expected to have a higher molecular fraction. The H2 electronic bands can, in principle, be seen in emission from the radiative cascade that would follow absorption of photons in the Lyman- and Werner bands. This would be most observable near regions with strong UV fluorescent excitation of H2 (see below). In particular, there have been observations of emission-lines from high-lying Werner bands in the 1150-1950 region accessible with IUE by Witt et al. 1989 [ApJ, 336, L21] towards a cold molecular cloud (the reflection nebula IC63), and by Schwartz et al. towards warm H2 in Herbig-Haro objects (shocks associated with jetlike outflows from young stellar objects).
Near-Infrared Vibrational Rotational Emission Lines
Within the electronic ground state of H2, all of the vibrational and rotational transitions are forbidden by the dipole selection rules. This means that the P- (J=1) and R- (J=+1) branches of H2 do not occur, but transitions in the O- (J=2), Q- (J=0), and S- (J=+2) branches can occur and are observed in the near- and mid-infrared. The first vibrational-rotational transition in the H2 ground state is the v=00S(0) transition, which is the J=20 transition at =28.2m. All other vibrational-rotational transitions in the ground state have increasing energy (and shorter ). Pure rotational lines (vu=vl) span the 3.4 to 28m region, while v=1 vibrational-rotational transitions have typical energies of ~0.5eV and are found in the 14m region clustering around 2m. Typical transition probabilities are ~107 108 sec1, so these lines are strongly forbidden.
ISO spectrum of Orion H2 Peak 1 from 2-40m showing a wealth of H2 vibrational and rotational emission lines [Rosenthal et al. 2000, A&A, 356, 705]
The notation used for these lines is: vu vl O(J l ); for J= 2 vu vl Q(J l ); for J=0 vu vl S(J l ); for J= + 2
V-13
Interstellar Molecules
For example, the 10S(1) emission line at =2.1218m in the Near-IR arises from a transition between the (v,J)=(1,3) upper level and the (v,J)=(0,1) lower level in the electronic ground state of H2. H2 has two spin isomers: Para-Hydrogen with the proton spins anti-parallel, corresponding to the even-numbered J states, and Ortho-Hydrogen with the proton spins parallel, corresponding to oddnumbered J states. The ortho-to-para ratio is 3.0 for thermalized level populations (75% ortho-H2 and 25% para-H2). Because only J=0,2 radiative transitions can occur, radiative transitions between the various rotational-vibrational levels in the ground state will preserve the ortho- or parastate of the molecules, and so one speaks of separate ortho and para energy ladders within the ground electronic state. Deviations from the predicted ortho-to-para ratio of 3 are primarily caused radiative de-excitation from the excited Lyman- or Werner-bands into ortho or para ladders in the ground state, and thus observations of different ortho-to-para ratios is taken as a sign of fluorescent excitation by UV photons. Because the rotational-vibrational states have typical transition energies of 0.5eV, they require excitation temperatures of ~1000 K. Since the kinetic temperature of typical cold molecular clouds is at most a few 10s of Kelvins, this means that you will only see these lines in regions that are unusually energetic. There are three basic mechanisms for exciting the NIR rotational-vibrational emission lines of H2 in nebulae: 1. Collisional Excitation: This primarily affects the low-J levels of H2. Hydrodynamic shocks with velocities of >6 km/sec are the principal cause of strong collisional excitation, but if the velocity is >25 km/sec, the shock will destroy (dissociate) H2. 2. UV Fluorescence: Radiative excitation of the Lyman and Werner bands by UV photons is followed by radiative de-excitation into excited states of the ground level which pumps the populations of the excited states. 3. Formation Excitation: Excitation of low-lying states as a by-product of molecular formation on grains due to the redistribution of binding energy between heating of the grain surface, kinetic energy of the ejected molecule, and internal excitation of the newly-formed H2. We shall consider each of these in turn.
Collisional Excitation
H2 molecules are subject to collisions in a variety of astrophysical environments, particularly in regions with hydrodynamic shocks such as jets from young stellar objects, stellar winds impinging upon interstellar clouds, or shocks in supernova remnants. Because the first (v=1) vibrational excited state of H2 lies ~0.5eV above the ground state (corresponding to E/k6000K), temperatures above ~1000K are required to collisionally excite the low-lying vibrational levels of H2. Because the dissociation rate of H2 increases very rapidly with temperature, molecular gas hotter than 4000-5000K will either rapidly dissociate into HI or cool dramatically (H2 dissociation is a very efficient cooling mechanism). As such, most collisional excitation of H2 will occur for molecular gas in a very narrow range of a few thousand Kelvins. Models of shocks show that the excitation of the low-J states responsible for the observed rotationalvibration levels occurs if the shock speed is faster than ~6 km/sec, with an excitation efficiency that scales roughly like n0v3, where n0 is the pre-shock density and v is the shock speed. At shock speeds >25 km/sec, the shock heating results in dissociated into atomic Hydrogen, but one may still get excitation due to re-formation behind the shock as the post-shock gas and dust grains cool off.
V-14
Interstellar Molecules
Because of the narrow range of excitation temperatures between the lower threshold and the upper dissociation/cooling threshold, collisionally excited H2 is characterized by strong lines of low vibrational levels (v<4), with rapidly decreasing level populations with increasing energy levels. In the high-density limit where collisional excitation and de-excitation will dominate, the relative H2 vibrational-rotational level populations nv,J are characterized by a thermalized Boltzmann population:
E exp v, J gJ kT where gJ is the statistical weight of the level, Ev,J is the energy of the upper level, and T is the kinetic temperature of the gas. For odd-numbered J-levels (ortho-H2), the statistical weight is:
nv , J
g J = 3(2 J + 1) while for the even-numbered J-levels (para-H2), the statistical weight is gJ = 2J +1 The standard analysis for the rotational-vibrational states of H2 is to create an Excitation Diagram that plots the emission line data as ln(nv,J/gJ) as a function of (Ev,J/k). The line strengths are usually expressed relative to a strong H2 line like the v=10S(1) 2.1218m line. If the level populations are thermalized, they will all lie along a straight-line locus in this diagram with a slope inversely proportional to the excitation temperature of the gas. An example excitation diagram for a collisionally excited region in the starburst galaxy NGC6240 is shown below.
Excitation diagram for near-IR H2 lines in NGC6240 (unpublished OSIRIS data)
If a range of temperatures occurs along the line of sight (e.g., your sight line integrates through the cooling region behind a shock), the diagram will show a smoothly curving locus of points, with lower energy levels lying along a steeply sloped (lower T) curve, with more highly-excited levels lying along a flatter-sloped curve because these levels are only excited at higher temperatures. An example of this is shown below for the bright near-IR H2 peak in the Orion Molecular Cloud (OMC Peak 1) from Everett, DePoy & Pogge 1995 [AJ, 110, 1295]. The curves are fits assuming a power-law cooling function responsible for the range of temperatures observed. As advertised, the low-lying levels have a steeper slope (lower excitation temperature) than the high-lying levels (shallower slope, hence higher excitation temperature).
V-15
Interstellar Molecules
Excitation Diagram for near-IR H2 emission lines in Orion H2 Peak 1 [Everett et al. 1995]
At lower densities, radiative de-excitation from excited rotational-vibrational states becomes important for determining the relative H2 level populations. Because each energy level has a different critical density, you to need specify both the density and temperature to estimate the equilibrium H2 populations. Such level populations will not be thermalized per se, and the temperature derived from the slope of the data in the excitation diagram is more correctly an excitation temperature rather than a reasonable approximation of the kinetic temperature. Which colliding species dominates (atoms or other molecules) determines the densities at which the observed relative level populations will deviate from the predicted high-density thermalized (nearLTE) populations. This is because each different collider has a different critical density associated with it. In purely molecular regions where H2-H2 collisions dominate, the critical densities are relatively high: 1056 cm3. In partially atomic gas, H0-H2 collisions begin to dominate when the atomic fraction is H0/H2>0.01 because the critical densities for H0-H2 collisions are about 2 order of magnitude lower than for H2-H2 collisions. In this case, the level populations will fall below the LTE predictions at densities of ~103 cm3. In either case, where a sufficient number of Near-IR H2 lines have been observed to make these subtle deviations evident, detailed level calculations are required to fully understand them (see Martin, Schwarz & Mandy 1996 ApJ, 461, 256, and Everett 1997, ApJ, 478, 246).
Fluorescent Excitation
In the interstellar medium, H2 in the electronic ground state can be excited into the first or second electronic excited states by absorption of a UV photon in the Lyman or Werner bands, respectively. Excitation is followed by radiative de-excitation back into the electronic ground state. On average, ~12% of the excited H2 molecules will de-excite into the dissociation bands (v>13) of the ground state, destroying the molecule (see below). The remaining, however, de-excite into bound levels of the ground electronic state which at low densities will cascade downwards into the v=0 level by radiative de-excitation through the forbidden rotational-vibrational transitions described before. At very high densities (>105 cm3, the critical densities of the rotational-vibrational states), the populations of these excited states can be redistributed by collisions before de-excitation into the v=0 level. Similarly at high UV flux densities multiple fluorescent excitations can redistribute the excited ground-level populations.
V-16
Interstellar Molecules
In addition to fluorescence by absorption of Far-UV continuum photons, the 12R(6) and 12P(5) transitions in the H2 Lyman band are in near-resonance with the HI Lyman- line at Doppler shifts of +15 and +100 km/sec respectively. Strong shocks can produce significant Lyman- emission, which in turn can fluorescently pump H2. This has been implicated as a possible source of warm H2 emission at the working surfaces of jets emerging from Young Stellar Objects (e.g., Wolfire & Knigl 1991 ApJ, 383, 205). Because the rotational state of the molecule changes very little during excitation and de-excitation (only J=0,2 transitions can occur), the rotational level populations will reflect the rotational state of the original gas being illuminated by UV radiation, whereas collisions can result in rotational mixing. In general, the spectral signatures of low-density fluorescent excitation of H2 are threefold: 1. Significant population of high-lying (v>4) vibrational levels that are rarely populated by collisions as they have sufficiently small Boltzmann factors that make these higher vibrational states relatively inaccessible. 2. Excitation temperatures for rotational states within a given vibrational level (rotation temperatures) are lower than the excitation temperatures among vibrational levels at a given rotational state (vibrational temperatures). In collisionally excited systems, rotational and vibrational temperatures are the same. The effect of this is to split the simple locus in the excitation diagram into separate curves by vibrational level. 3. The populations of even-numbered rotational states (para-H2) are enhanced relative to oddnumbered states (ortho-H2), in the sense that the observed ortho-to-para ratio will be smaller than the value of 3 expected for a thermalized population. An example of an excitation diagram for fluorescently-excited H2 gas in the reflection nebula NGC 7023 from Martini et al. 1997 [ApJ, 484, 296] is shown below.
Excitation diagram of fluorescent H2 lines in NGC 7023 (adapted from Martini et al. 1997)
The effects of fluorescent excitation are most dramatic among the high vibrational levels (v=5, 6, and 7) that occur primarily in the 11.8m region (J and H-bands) and are unlikely to be significantly populated by collisions. The spectral signature of fluorescently excited H2 emission has been observed in reflection nebulae (Hayashi et al. 1985 MNRAS 215, 31P, Sellgren 1986 ApJ, 305, 399; Martini, Sellgren & Hora 1997, ApJ, 484, 296; Martini, Sellgren, & DePoy 1997, ApJ, 484, 296), and planetary nebulae (Dinerstein et al. 1988 ApJ 327, L27).
V-17
Interstellar Molecules
The signatures of fluorescent excitation, however, can be effectively erased at high densities when collisions start to become important. If the density is high enough (>1045 cm3), collisions can thermalize the level populations, resulting in excitation diagrams that more closely resemble shockheated H2, even when UV fluorescence is important. The effect of collisions is to quickly de-populate the high vibrational levels whose lines are usually an unambiguous signature of fluorescent excitation. The converse, however, does not occur: there is no convincing combination of effects that will cause a collisionally excited region to emit like a fluorescently excitation region. In any case, the detailed predictions of fluorescent excitation models depend sensitively on the density, temperature, UV radiation field intensity & spectrum, H2 formation/reformation processes etc. Classic fluorescent models are those of Black & van Dishoeck (1987 ApJ, 322, 412) and Draine & Bertoldi (1996, ApJ, 468, 269). These are the ones that are most often compared with observations.
Formation Excitation
H2 molecules form most efficiently on the surfaces of dust grains. Upon formation, the 4.5eV binding energy must be distributed into kinetic energy (after breaking the bond with the grain surface), heating of the dust grain at the formation site, and internal energy in the form of excitation of rotationalvibrational states within the electronic ground state. The exact distribution of energy depends on the details of formation that are fundamentally unknown, and so formation excitation is difficult to assess observationally. An example is higher-than-expected excitation temperatures in H2 Lyman-Werner bands observed in absorption towards stars (e.g., Spitzer & Zweibel 1974, ApJ, 191, L127). Some models have tried to include formation excitation in an ad-hoc way. For example the models of Black & van Dishoeck (1987) considered three different distribution functions in which (1) 1.5eV was distributed among the rotational-vibrational levels following the Boltzmann distribution, (2) all H2 molecules from in the v=14, J=0,1 levels, and (3) all H2 molecules form in the v=6 level with low-J states. These are then used as inputs in computing the emergent spectrum (see Black & van Dishoeck 1987 and Le Bourlot et al. 1995 ApJ, 449, 178 for examples).
Formation of H2
There are two primary ways that H2 can be formed in the gas-phase. The first is Direct Radiative Attachment in which two H atoms collide to form H2 followed by radiation of the excess 4.5eV binding energy as photons:
H 0 + H 0 H2 +
After the H0-H0 collision the binding energy is converted into internal energy (excitation of rotationalvibrational levels of the ground state) in newly-formed H2 molecule. Because these transitions are highly forbidden, with small Einstein A coefficients, radiation of the binding energy very inefficient, and the molecule is much more likely to dissociate before it relaxes. The rate coefficient for radiative attachment is <1023 cm3 s1, so this mechanism is much too slow and inefficient to form H2 in sufficient quantities to be important in the ISM. Three-body processes, in which three H0 atoms collide and the third H0 atom carries off the excess binding energy, are similarly unlikely at ISM densities, though they have been proposed as a way to form H2 in protoplanetary disks. The second gas-phase mechanism is Associative Detachment, which was first proposed by McDowell [1961, The Observatory, 81, 240] and given its classic treatment by Dalgarno & McCray [1973, ApJ, 181, 95]. H 0 + e H + h
H + H 0 H 2 + e
V-18
Interstellar Molecules
The first reaction, Radiative Association, is slow, having a temperature-dependent rate coefficient of ~1018 T cm3 s1. The second reaction, Associative Detachment, is much faster with a nearly temperature-independent rate coefficient of ~1.3109 cm3 s1, though this step will have to compete with photodissociation of H which has a rate of ~2.4107 s1 for diffuse clouds exposed to the ISRF. The combined volumetric H2 formation rate for this mechanism is approximately
dnH 2 dt
xe T n n cm 3 s 1 7.5 1023 4 H H 0 100K 1.4 10
Overall, this mechanism is also very inefficient at forming H2 in the ISM, especially at low temperatures (though it is about an order of magnitude faster than direct radiative attachment). While too slow to explain the observed abundances of H2 in the present-day ISM, it may be important in the early universe when no metals, and hence no dust grains, were present, but the gas was hotter (larger T) and there were copious electrons present from partial ionization of H0 (larger xe) The currently favored H2 formation mechanism is Grain-Surface Catalysis, first described by McCrae & McNally [1960, MNRAS, 121, 238] and given its classic treatment by Gould & Salpeter [1963, ApJ, 138, 393]. An H atom colliding with a dust grain will have a certain probability of sticking to the surface, provided that it is not moving too fast (e.g., as in a hot HI gas) and that the grains themselves are not too hot, which makes the surfaces energetically less sticky. Once the H atom has been adsorbed by the grain surface, it will migrate around until it reaches a site on the grain surface where it is more tightly bound by either chemisorption (i.e., bound by valence forces with surface materials) or physisorption (i.e., bound by intermolecular van der Waals forces). Such sites act as sinks for H atoms. It is while trapped in this site that the atom is likely to encounter another H atom, and the two can react to form H2, distributing the 4.5eV of binding energy into (1) heating of the dust grain, (2) breaking the activation barrier at the formation site and ejecting the newly-formed H2 molecule from the surface (desorption), and (3) internal vibrational energy (excited ground states) that is later radiated away by the molecule as rotational-vibrational emission lines. The volumetric rate of dust-catalyzed H2 formation is given by:
dnH 2 = Rd nH nH 0 dt where nH is the total density of hydrogen of all forms (nH=nH0+2nH2), nH0 is the density of neutral atomic hydrogen, and R is a rate coefficient with units of cm3 s1 that depends on the temperature of the atoms and grains and the fraction of atoms that can adhere to the grain:
1 8kT Rd = m 2 H
1/2
da n
1 dngr 2 a (a) H da
Here (a) is the probability of sticking to a grain of size a. In general, it appears that all H atoms will stick to grains if their speeds are <2km/sec, provided that the grain temperatures are between ~20100K. If the grains are too hot, the adsorbed atoms will evaporate off the grain surface before they can find another H atom and form H2. From observations of the fraction of H2 towards nearby stars using Far-UV measurements of the Lyman-Werner bands, the H2 formation rate coefficient is estimated to be Rd131017 cm3 s1, implying a mean sticking fraction of 0.06 for typical interstellar conditions and the observed dust/gas ratio. Draine & Bertoldi [1996, ApJ, 468, 269] quote Rd=61018T1/2 cm3 s1. Thus while very small grains have most the total grain surface area, they are expected to be less effective as H2
V-19
Interstellar Molecules
formation sites, and most molecular formation thus occurs on larger grains. In some sense this is not surprising, as the smaller the grain becomes, the longer the timescale between H-atom collisions. The H-atom impact timescale
a tHI 810 s 0.01m
4
2
nH 30 cm3
1
T 100K
1/ 2
For typical interstellar conditions, tHI is of order the photon absorption time. As we saw in the previous section on Dust, the absorption of a single photon by a small grain can cause rapid stochastic heating of the grain. Thus in the time it takes 2 H atoms to collide with and stick to a small grain it is very likely that a photon will be absorbed with enough energy to kick off the one or both of the hydrogen atoms, effectively suppressing H2 formation on tiny grains. Such photons have a less dramatic effect on the temperatures of large grains, and so photosuppression of H2 formation will be less effective. The corresponding timescale for H2 formation on grains, tH2 is:
t H 2 109 nH1 years
For typical interstellar cold interstellar clouds where densities are of order 104 cm3 it is possible to convert most of the HI into H2 in as little as 105 years. Interstellar shocks may speed up this process by making the gas warmer and denser, provided the shocks are not so strong as to destroy the grains or the newly formed molecules. Some other important papers on H2 formation on grains are Hollenbach & Salpeter 1971 [ApJ, 163, 155] and Hollenbach & McKee 1979 [ApJS, 41, 555], and discussion in Draine & Bertoldi 1996 [ApJ, 468, 269].
Destruction of H2 Molecules
H2 molecules can be destroyed collisions and by UV photons (photodissociation). Collisional dissociation occurs at temperatures >4000 K, when collisions with HI can excite H2 into the unbound v>13 vibrational states. Photodissociation of H2 occurs when UV photons (either from a nearby star of the ISRF) are absorbed with sufficient energy to unbind the molecule. There are two possible processes for photodissociation:
Direct Photodissociation from the electronic ground state: An H2 molecule in the X1g+ ground state should dissociate if it absorbs a photon with an energy >14.7eV (<850). The problem with this mechanism is that there are two processes competing for these photons. The first is HI ionization (threshold of 13.6eV, =912), which has a higher cross-section, so such photons are more likely to photoionize HI than dissociate H2. Second, the threshold for photoionization of H2 to H2+ out of the ground state is h>15.5eV (<804), which also has a larger cross-section than photodissociation. Thus while direct photodissociation can occur, it is very inefficient. UV Fluorescent Photodissociation This is a two-step process in which absorption of Lyman- or Werner-band UV photons is followed by radiative de-excitation into ground state vibrational continuum bands which have vibration quantum numbers v>13. These high vibrational levels are unbound and the molecule dissociates, converting binding energy into the kinetic energy of the now free H0 atoms. On average ~12% of all radiative decays following absorption in the Lyman- and Werner-bands
V-20
Interstellar Molecules
results in fluorescent dissociation of H2. This is the dominant mechanism for H2 photodissociation in the ISM. Classic papers describing these H2 destruction are Dalgarno & Roberge [1979, ApJ, 233, L25 for collisional dissociation] and Black & Dalgarno [1976, ApJ, 203, 132 for photodissociation]. The rate of photodissociation via Lyman/Werner band fluorescence is
diss = 4 J k d
where 4J is the strength of the radiation field (e.g., the ISRF in the diffuse ISM), k is the fraction of Lyman/Werner band excitations that are followed by de-excitation into the vibrational continuum band and lead to dissociation of H2 (k0.12 for the ISRF), and is the cross-section for LymanWerner fluorescence
= i i
i
where the sum is over all Lyman/Werner bands with h <13.6eV (i.e., absorptions arising from photons less energetic than the HI ionization potential). The rate of H2 destruction via this process is dnH 2 dt
= diss nH 2
In the diffuse ISM, equilibrium between grain formation and fluorescent photodissociation occurs when Rd nH nH 0 diss nH 2 For typical values of the rate coefficients in the diffuse ISM, diss5.51011 s1, and so the fraction of H2 relative to Hydrogen of all forms in the diffuse ISM is roughly nH 2 nH
1.1 107 T 1/2 nH
In the diffuse ISM, the time to achieve equilibrium should be of order 1/diss, which is ~600 years for typical conditions. We thus expect that in the diffuse ISM, the fractional abundance of H2 that we observe along virtually all sightlines is in equilibrium.
Self-Shielding
At H2 column densities >1014 cm2 the Lyman/Werner UV absorption line bands start to become optically thick, and H2 becomes self-shielding in the sense that all of the photons that could lead to UV fluorescent photodissociation are absorbed by H2 in the outer layers of the cloud, effectively shielding the H2 within it. The rate of H2 photodissociation thus depends on the abundance and state of excitation of the H2 as you go into the cloud. Self-shielding occurs in diffuse clouds exposed to the ISRF or in dense clouds in proximity to sources of UV photons (e.g., reflection nebulae around earlytype stars or HII regions and PNe). Dust can also absorb UV photons, further limiting the process, but as we will see it dominates only when the local UV radiation field is unusually intense relative to the density of the cloud. Self-shielding makes H2 a very robust molecule, allowing it to exist even in the UV radiation field of the diffuse ISM.
V-21
Interstellar Molecules
The H2 photodissociation rate in the presence of dust opacity and H2 self-shielding is
diss = 0e
gr
f shield ( N H 2 )
where 0 is the dissociation rate at the cloud surface, gr is the dust grain optical depth, and fshield is the self-shielding factor, which Draine & Bertoldi [1996, ApJ, 468, 269] estimated as
1 = N H 3/4 2 N0 for N H 2 < 1014 cm 2 for 1014 N H 2 < 1021 cm 2
f shield
Below the threshold column density of N0=1014 cm2 there is no self-shielding, whereas above it selfshielding scales as the H2 column density to the 3/4 power. The strong dependence on column density means we expect there to be a very sharp-edged transition between atomic and molecular gas regions as we go into a molecular cloud, much in the same way that we saw an abrupt transition between ionized and neutral gas in HII regions. This can be seen as follows. In steady state, H2 destruction is balanced by H2 formation
diss nH = Rd nH nH
2
0
Assuming low dust opacity and recalling that nH=nH0+2nH2, the steady-state density of H2 is
nH 2
3/4 0 N0 = nH + 2 3/4 Rd nH N H 2 1
Since column density (N) and total density (n) are related by dN=n ds, we can rewrite this to solve for the approximate H2 column density: N H2 and the H2 abundance, XH2 X H2 = dN H 2 dN H R n N = 4 d H H 4 0 N 0
4 3
R n N = d H H N0 4 0 N 0
4
4
The result is that there is a very sharp transition between neutral atomic hydrogen (H0) and molecular hydrogen (H2) going into a region of steady-state photodissociation because of the effect of selfshielding. The mid-point of the transition will occur when the gas is half atomic and half molecular which occurs when XH2=1/4. This transition point occurs at a column density of N diss
G 3.7 10 0 nH
22
4 /3
cm 2
where G0 is the mean intensity of the ISRF and taking Rd=31017 cm3 s1, the upper end of the range quoted above. Dust opacity starts to become important when A11001 hence NH21020 cm2. This means that the onset of H2 self-shielding will determine the location of the H2/H0 transition for (G0/n)<0.02 cm3, which is generally true of molecular clouds illuminated by the ISRF. Near OB starforming regions or reflection nebulae in proximity to UV-bright stars, however, (G0/n)~1 cm3, and dust opacity, not self-shielding, determines the location of the molecular/atomic transition which works out to be a depth where AV=2 going into the molecular cloud.
V-22
Interstellar Molecules
V-3 Observed Properties of Molecular Clouds
In this section we will briefly review some of the properties of giant molecular clouds as derived primarily from observations of tracer molecules like CO. This is a research area that has become increasingly active as new and better millimeter techniques; especially millimeter interferometry and sub-millimeter imaging (e.g., SCUBA) have come into use. A thorough review of the current literature would be beyond the scope of this course, but this should give you an idea of some of the more common results quoted in recent and previous papers.
Cloud Structure
Local density estimates using line ratios often give larger densities than global mean densities found by averaging the observed molecular column densities along the line of sight. The interpretation of this is that the clouds are very clumpy, with the dense cores having typical sizes of <1 pc or smaller, and densities >106 cm3. The overall cloud itself extends for 320 pc on average, with a mean density of 1034 cm3. Some molecular clouds (most?) have a number of discernable cores. These are often detected as sources of molecular lines with high critical densities (e.g., CS), while the general cloud is mapped using lines of lower critical density (chiefly CO). Within the galaxies, molecular clouds are most often seen organized into complexes with sizes from 20 to 100pc, and overall H2 masses of 1046 Msun. The distinction between clouds and complexes in terms of sizes and masses is somewhat artificial. A more precise statement is that we see a wide range of structures, from single small clouds to large complexes of clouds, with many complexes arrayed along the spiral arms of the Galaxy.
Cloud Masses
One way to estimate the mass of a molecular cloud is to integrate the column density across the face of the cloud by mapping it with millimeter interferometry. A common alternative that is used is to assume that the cloud is only supported against its own gravity by internal thermal pressure. In this case the characteristic length scale within the cloud is the Jeans Length:
kT RJ 2 Gn
1
2
Assuming that the clouds are spherical, the mass will be the Jeans Mass: 4 3 RJ C 18M sunT 1.5 n 0.5 3 where T is the kinetic temperature in K, and n is the density in cm3. If the cloud is supported against gravity by random turbulent velocities with a characteristic velocity dispersion of v, you can define a Virial Mass for the cloud R M vir v 2 G Where R is the radius of the cloud, and v is the velocity dispersion of an optically thin emission line (e.g., the 13CO J=10 line). The assumption here is that the observed velocity dispersion is just the amount needed to virialize the cloud. Computation of a virial mass requires that you can resolve the cloud (measure R). A more sophisticated analysis assumes that the cloud has a radial density profile, rathe...