A two-component LVG model of the molecular gas in M82 is consistent with the measured 12CO and 13CO spectra up to J=7-6, and matches the large observed
J=2-1/J=1-0 and J=2-1/J=3-2 ratios. Results of this model suggest that the nu- clear region contains a large number of small, cool molecular clouds, and a small number of large, warm, low-density clouds.
I have demonstrated an approach to the evaluation of excitation models that pro- vides important insight into how well the model parameters and various quantities derived from them are constrained by the measured data. This is accomplished by computing a likelihood density curve for each of the parameters and derived quanti- ties. I have also demonstrated how prior knowledge and/or physical constraints can be incorporated into this evaluation. We find that a range of conditions is consistent with the measured M82 data. This is to be expected for a complicated region like a galactic nucleus; simple excitation models cannot be expected to determine precise results. On the other hand, simple models can be used to exclude large regions of parameter space.
The results of this analysis show that the most precisely determined quantity is the 13CO beam-averaged column density, with a value of about 4.5×1016 cm−2 in both lobes. The beam-averaged column density of 12CO was determined to within a factor of a few to be 1018.2 cm−2 in both lobes. The area filling factors were found
to be near unity for the cool molecular gas, and around 0.03 for the warm molecular gas.
Median likelihood estimates of the isotopomer abundance ratio12CO/13CO in the
NE and SW lobes of M82 are 40 and 30, respectively. These estimates lie between the value of 25 for Galactic center clouds determined by G¨usten (1989) and values
ranging from 50 to 100 for M82 determined by Mao et al. (2000) and Weiß et al. (2001).
Although the temperature of the warm molecular gas in the nucleus of M82 was not well constrained (no upper limit to the temperature was found), it is likely that over half of the total molecular mass is warmer than measured dust temperatures of 48 K.
The density of warm molecular gas in the nuclear region of M82 is low, with median likelihood estimates around 103 cm−3. Cool molecular gas appears to be
typically more dense than warm molecular gas. Median likelihood estimates for the warm gas and cool gas pressures in the two lobes range from 104.4 to 105.2 K cm−3.
The 12CO J=6-5 to 12CO J=2-1 line ratio is at least five times larger in the SW
lobe of M82 than in the galactic center, indicating that the proportion of molecular gas that is warm is significantly larger in the nuclear region of M82 than in the Galactic center.
Chapter 7
Concluding Remarks
This thesis described one small part of the complex web of research necessary to understand the molecular gas feeding star formation in starburst galaxies. Starting from electrical models, software was written to evaluate and optimize submillimeter heterodyne receiver designs. Using this software, a 690 GHz receiver was designed to observe J=6→5 rotational emission spectra of carbon monoxide in extragalactic sources. The receiver was used at the Caltech Submillimeter Observatory to map car- bon monoxide emission toward the nearby starburst galaxy M82. These observations, along with measurements of nine other carbon monoxide emission lines were used to study the physical conditions of the molecular gas in the nucleus of the galaxy.
Research is an ongoing process. Each hard-earned answer suggests new questions, and each incremental improvement of a technology becomes a starting point for the next generation of advances. I will thus conclude this thesis by looking forward and saying a few words about where this research might lead.
Submillimeter receiver technology is developing at a rapid pace. Several new receivers are currently being designed and fabricated by the Caltech Submillimeter Wave Astrophysics group that will provide a major boost in capability over the cur- rent generation of instruments. Fast, accurate computer simulation and optimization tools including SuperMix are enabling this rapid development. SuperMix works, but there is still room for continued improvement. In particular, due to the lack of a comprehensive user’s guide, only a few people outside of Caltech have been able to take advantage of the possibilities that this software offers. Expanding to a larger pool of users would ensure that the full potential of this new software tool is realized. The long-term usability would be further enhanced if the users begin to add new components to the library as receiver technology evolves to utilize new and different materials and techniques. We have provided a valuable tool to the receiver develop- ment community, but the task still remains to make it more generally accessible.
Improving the performance of the receiver would be straightforward. First, the low-noise IF amplifier could be retuned for lower noise and flatter gain. The mixer chip could be redesigned to lower its capacitance; a lower capacitance would allow a better impedance match to the IF amplifier, with the result being lower, flatter noise across the IF passband. Taking the details of the circuit geometry around the SIS mixer chip into account, it should be possible to keep the gain flat and the noise low across the entire 4 GHz of IF bandwidth. The noise could be further reduced by carefully minimizing optics losses.
After the performance of the single-mixer receiver has been fully optimized, two mixers could be combined as a correlation receiver to eliminate the need for beam switching (Blum 1959; Faris 1967; Predmore et al. 1985). This system would in- crease the on-source integration time by a factor of 2.5 to 3, and the subtraction of correlated sky noise would be significantly improved. Simultaneously observing both polarizations of the incoming radiation could further improve the sensitivity of the receiver by a factor of √2.
The next step for the study of molecular gas in M82 would be to observe 13CO
emission in the J=4→3 and J=6→5 transitions to improve constraints on the column density, isotopomer abundance ratio, and molecular gas mass. The observations and excitation analysis should be repeated for many different starburst galaxies. The results of such a survey could then be studied to find trends that apply to starburst galaxies in general.
The Atacama Large Millimeter Array (ALMA) is scheduled to begin operation as early as 2006 (ALMA web page). The site is located at an altitude of 16,400 feet in the dry Andes mountains of northern Chile, and has better weather for submillimeter astronomy than Mauna Kea. The zenith transmission in the 650 GHz window is higher than 40% over a quarter of the time. With 64 12 meter dishes, the total col- lecting area will be 85 times larger than the CSO’s 10.4 meter antenna, and baselines up to 14 km will enable spatial resolutions as high as 0.0100, corresponding to 0.2 pc at the distance of M82.
enable the detection and mapping of high-J lines of optically thin CO isotopomers. Excitation analyses similar to those presented in Chapter 6 could be repeated at high resolution. Such observations will greatly improve our understanding of the interstellar medium of starburst galaxies.
Appendix A
The SuperMix Circuit
Connection Algorithm
An overview of composite circuits in SuperMix was given in Section 2.4, which de- scribed how SuperMix calculates composite circuits one connection at a time using the method of subnet growth. A binary tree of connections is formed, with the com- posite circuit forming the base of the tree, the user-defined circuit elements the tips of the branches, and each connection a branch of the tree; see Figure 2.3.
An example was presented in Section 2.9.1 of how a composite circuit can be simulated. The circuit is represented as an object of type class circuit. The components of the circuit are connected by calling the connect member function of
class circuit. The order of the unconnected ports in the composite circuit are
specified with the add portmember function. Once the circuit has been specified by the appropriate calls to these functions, the circuit behaves just like any other linear device in SuperMix, and can itself be used in other, more complex circuits.
In this appendix, we will first examine two classes used extensively in building composite circuits, i.e., class port and class connection. We will then examine the workings of class circuit in more detail, listing some of its more important member data structures and presenting algorithms for several of its member functions.
A.1
Identifying Ports
Every device in SuperMix is automatically assigned a unique number called a “device ID.” Each device, in turn, assigns an integer to each of its ports. Thus, any port within a simulation can be uniquely identified with a (device ID, port number) pair. SuperMix defines aclass portdata type to hold this pair. A connection can then be specified with a pair of class portobjects identifying the two ports to be connected.