Your PhD in 1976 dealt with hydroxyl masers in the envelopes of long period variable stars. What motivated that subject and how did it lead into your later work on star formation?
When I started my graduate work at Caltech with Duane Muhleman, he asked me to help on a radio wavelength version of the famous gravitational light-bending experiment of 1919. This involved VLBI observations of two quasars, 3C273 and 3C279, as the Sun moved between them and deflected the radio waves. This was early in the development of VLBI and these observations failed owing to various technical problems. Even so, the technique left me very excited about the possibility of obtaining milli-arcsecond resolution. While radio interferometers had been making images, at that time VLBI data could not be phase calibrated, and analyses only involved model fitting of amplitudes. Since spectral-line VLBI data can be fully calibrated against one maser spot, in my PhD thesis I constructed the first synthesis maps from VLBI data.
I became interested in advancing the technique of VLBI synthesis mapping as a postdoctoral fellow at the Center for Astrophysics. Jim Moran and I mounted a large “campaign” to get 8 antennas across the continental U.S. to image the hydroxyl masers in W3OH. In 1976 this was a very challenging project, not only to get all those antennas working together, but also to correlate and analyze the “huge amount” of interferometric data (64 spectral channels for 28 baselines time averaged to 10 seconds). While by today's standards such a data set would be trivial, it took me an entire summer keeping the NRAO IBM360 computer running 24/7 to make the synthesis images. These maps, published in 1980, provided deep insights into the maser amplification process, the structure of molecular gas surrounding compact HII regions, and measurements of strong magnetic fields (several milli-Gauss) in the gas from fully-resolved Zeeman split lines.
So, to answer your question, it was learning the technique of spectral-line VLBI that lead me to work on star formation, and later on Galactic structure and black holes. Regarding this, I'd like to make an “editorial comment'' about doing research. Broadly speaking there are two approaches to doing research in astrophysics: 1) starting with an interesting question and then looking into how it can be solved, or 2) starting with expertise in a technique (observational or numerical) and looking into which questions it can address. Many favor approach 1) and call themselves multi-wavelength astronomers. I've mostly favored approach 2) as this leads to technical advances which offer unique answers to astrophysical questions. Both approaches are valid and should be encouraged.
In 1985 you and Paul Ho published a paper on the cometary appearance of the HII region G34.3+0.2. What motivated this discovery?
The VLBI synthesis images of W3OH just mentioned firmly established that the masers were tightly distributed around and in front of an ultra-compact HII region. This very interesting case study motivated Paul Ho and I to use the newly constructed VLA to image a sample of ultra-compact HII regions with OH masers. In those days one travelled to the VLA site in order to monitor observations as well as to analyze the data with special software on a DEC10 computer. While most of the compact HII regions displayed quasi-spherical structure, G34.3+0.2 stood out dramatically with a comet-like appearance. The contour map in our paper does not do justice to the stunning image that appeared in grey tones on our computer screen at the VLA. This was the first clear member of a significant sub-class (perhaps 20%) of compact HII regions, which we called “cometary” HII regions.'' Their morphology is likely owing to supersonic motion of the exciting O-type star with respect to surrounding material.
A few years later you and your collaborators used H2O masers to determine the distance to the center of our Galaxy. This is a subject that you have focused on for many years, with an Annual Review of Astronomy and Astrophysics article in 1993 and a more recent parallax determination of Sgr B2. How did this begin and how well do we now know this important parameter?
In the late 1970s, Reinhard Genzel suggested that the proper motions of water masers in star forming regions could be measured and reveal their 3-dimensional kinematics. Also by comparing radial speeds with proper motions one could infer an approximate source distance. This lead to direct distance estimates for Orion-KL and W51. While these distances had relatively low accuracy, we realized the potential. With a series of VLBI observations of the water masers toward Sgr B2, which is very close to the Galactic Center, we found R0 = 7.1 +/- 1.5 kpc. I have been working to improve the accuracy of R0 ever since.
In 1989, I was invited to give a review talk on the distance to the Galactic Center for IAU Symposium 136. In that review, I combined many published estimates of R0 to obtain a “best value.” My approach was to deal with the correlations among estimates that relied on the same fundamental calibration. So, for example, all published R0 values that used Cepheid distances likely had a common systematic uncertainty owing to the assumed P-L calibration. Along the way, when I plotted estimates of R0 versus date of publication, it became clear that there was a clear downward drift over time, likely a “bandwagon effect” wherein the editing of astronomical data is affected by “current wisdom” as to the value for R0.
Later, I was invited to expand the analysis for a paper in Annual Reviews. I attempted to homogenize the estimates of R0 by adjusting them to a consistent set of calibrations and gave a best estimate of R0 = 8.0 +/- 0.5 kpc. In the last paragraph of the paper, I looked to the future and pointed out that a trigonometric parallax for Sgr A* was possible and that by about the year 2000 “we can look forward to knowing the value of R0 to better than 3% uncertainty.” That was a bit optimistic and only recently has this level of accuracy been achieved: 8.15 +/- 0.15 kpc (Reid+ 2019), 7.96 +/- 0.07 kpc (Do+ 2019), and 8.25 +/- 0.05 kpc (Gravity Collaboration 2020).
You have also written several papers on the proper motions of Sagittarius A*. How has this subject evolved up to today and what have you learned?
As early as 1990 I had proposed for observing time on the partially constructed VLBA to measure changes in the apparent motion of Sgr A* over time relative to distant quasars. The proposal mentioned several measurements that would be interesting:
1) the angular motion of the Sun in its Galactic orbit (about 6 mas/yr),
2) a possible motion of Sgr A* with respect to the dynamical center of the Galaxy, and
3) the apparent change in position over a year's time caused by the Earth's orbit around the Sun (trigonometric parallax of roughly 0.1 mas).
We learned a lot from early observations, including that we should observe at the highest possible frequency to limit effects of scattering from clumps of interstellar electrons and that summer evenings were particularly phase unstable for most of the sites. In order to improve astrometric accuracy, we developed a technique to measure and remove the effects of changes in atmospheric delay on hourly timescales. This allowed us to calibrate, image, and measure relative positions with better than ~0.1 mas accuracy. However, even with this accuracy we could not measure a high-quality trigonometric parallax within a reasonable amount of observing time.
While a parallax measurement was not feasible, we could measure the proper motion and address the first two effects mentioned above. (At the time, Don Backer and Dick Sramek were measuring the proper motion of Sgr A* with the VLA, although I was unaware of their program as they had not published any results.) Our latest results (Reid & Brunthaler 2020) with data spanning 18 years show that Sgr A* appears to move along the Galactic plane at -6.411 mas/yr and out of the plane at -0.219 mas/yr, with an incredibly small uncertainty of +/-0.008 mas/yr. This apparent motion can be entirely explained by our changing vantage point as the Sun orbits the Galaxy.
Any intrinsic motion of Sgr A* relative to the dynamical center of the Galaxy is on the order of 1 km/s. The implications of this are profound. While stars at the Galactic center are seen to move at thousands of km/s, Sgr A* is anchored there and so must be extremely massive. This requires the radiative source, Sgr A*, to have a significant fraction of the 4x106 solar masses inferred by the orbiting stars seen in the IR by Genzel's and Ghez' groups, for which the 2020 Nobel Prize in Physics was awarded. Their case that Sgr A* must be a supermassive black hole was extremely strong; adding the lack of motion and near Schwarzschild-radius size from radio observations puts the “final nails in the coffin.”
The radio source Orion-I in the Kleinman-Low nebula is a mysterious object. You have observed it with the VLA, what did you conclude?
In 1990, I started to develop a technique to calibrate radio interferometric data where one could use strong maser line emission to phase calibrate and image nearby continuum sources. At that time 43 GHz SiO maser emission toward the Orion-KL star forming region was thought to be from an enimagatic IR source, IRc2. Karl Menten and I realized that the technique used to register maser and continuum emission could tell us a lot about IRc2, which had been suggested to power the KL nebula and drive its strong H2O masers. So in 1994 we observed this region with the VLA, which at the time had 9 antennas outfitted with 40-50 GHz receivers. Our results were very informative. We found that the SiO masers were coincident with a radio continuum source, Source-I, and that this source was offset by a few arcseconds from the peak of IRc2. Since SiO masers operate at physical temperatures of nearly 1000 K, this required Source-I to be reasonably luminous (~104 LSun). However, it also suggested that IRc2 was not a dominant energy source and the infrared came from reprocessed stellar photons in surrounding clumpy dust.
Our estimate of the luminosity of Source-I was most likely well below the total luminosity of ~105 LSun of the KL nebula. Where could most of that additional energy come from? In our radio maps we found another source about 3 arcsec south-west of Source-I and coincident with IR source “n”. Since that IR source had not been prominent on IR maps at wavelengths longer than 3.8 microns, it had not been considered as contributing significantly to the KL luminosity. When we found that source “n” was not only at the center of a dense ``hot core seen in ammonia, but also consistent with being at the center of expansion of the powerful H2O maser outflow, we considered it a good bet to be a dominant energy source for the KL region.
In late 2000, we re-observed Source-I with the VLA, now with improved receivers on all antennas. We saw a near edge-on disk-like structure in the 43 GHz continuum, which was precisely centered on the SiO masers. VLBA observations led by Lincoln Greenhill a couple of years earlier had shown that the SiO masers had an “X-like” structure, and it soon became clear that we were seeing an ionized accretion disk from a massive young star. Later, Greenhill's SiO maser proper motions clearly indicated that rotating and outflowing molecular gas was emanating from the disk, and not from the star.
In 2009 you, Karl Menten, and Zheng Xing-Wu along with a team of collaborators published no less than 7 papers on trigonometric parallaxes of massive star-forming regions. Since then many more papers have appeared in this series. Please tell us about the origin of this project and give some highlights.
While studying what limited the astrometric accuracy for the Sgr A* proper motion measurements, I found that a major source of error in relative position measurements for VLBI at frequencies above about 10 GHz was unmodeled propagation delays through the atmosphere caused mostly by water vapor. By rapidly observing about a dozen quasars with different source elevations across the VLBI array we could monitor and correct for the mismodeled delays by about a factor of 5. This suggested that we should be able to achieve astrometric accuracies in the 10s of micro-arcseconds for sources less affected by interstellar scattering (ie, allowing the use of much longer baselines) than Sgr A*. So in 2003 we began a series of VLBA observations of the 12 GHz methanol masers in our favorite source, W3OH, with the goal of measuring a trigonometric parallax. We measured a parallax of 0.512 +/- 0.010 mas (Xu+2006), which showed that the Perseus spiral arm was only 2 kpc from us in this direction -- about a factor of two nearer than implied by kinematic distances.
With the +/- 0.01 mas uncertainty we achieved, one can obtain a 10% accurate measurement for a source at a distance of 10 kpc, for example, well past the Galactic center! This lead us to propose an extremely large project designed to measure hundreds of parallax distances to star forming regions. We named this the Bar and Spiral Structure Legacy (BeSSeL) Survey, after Friedrich Bessel who measured the first accurate stellar parallax, and were given 3,500 hours of time on the VLBA over 5 years. With a large group of students and postdocs associated with Zheng Xing-Wu at Nanjing University and Karl Menten at the MPIfR, we were able to analyze an enormous amount of data and measure distances to well over 100 massive young stars.
One of the early highlights of the BeSSeL Survey was that a large number of sources between about 70o and 90o Galactic longitude, which we expected to be in the Perseus spiral arm, were much nearer and populated the Local arm. The Local arm hosts multiple star formation sites at different distances, which are projected along our line-of-sight and collectively called the “Cygnus-X” region. This changed our perception of the “under-appreciated” Local arm, and revealed that the Perseus arm had far less star formation than previously thought.
In 2014 you and your collaborators published a hugely influential paper on what parallaxes of high-mass star-forming regions can tell us about the structure and kinematics of the Milky Way. Another study followed in 2019. What are the key results?
These papers combined results from the BeSSeL Survey and the Japanese VERA project yielding about 200 parallaxes to high-mass star-forming regions. The major results include mapping the spiral structure of the Milky Way between Galactic longitudes of 0o to 240o. Most of the sources were within about 10 kpc of the Sun; however Alberto Sanna recently measured a parallax for a source on the far side of the Galactic Center at a distance of 20 kpc! This proved important for us to be able to extrapolate our spiral arm model through the fourth quadrant and behind the Center. We find strong evidence that the Milky Way is a four-armed spiral, with some extra arm segments (such as the Local arm).
With parallaxes, proper motions, and radial velocities, we have full 6-dimensional phase-space information for massive star-forming regions. Modeling the measured 3-dimensional velocities with only a handful of parameters, we find R0=8.15+/-0.15 kpc, a rotation curve that peaks at 237 km/s at a Galactocentric radius of 7 kpc and falls very slowly at larger radii. It is important to note that our rotation curve is based on 3-dimensional velocities, “gold standard” distances, and an internally estimated Theta0 value.
Interestingly, we find that the Sun is only 6 pc above the Galactic plane. Here we define the plane based on the average z-height of massive star-forming regions within 7 kpc of the center. These regions have a very tight z-distribution with a scale height of only 19 pc. Our result nicely explains why Sgr A*, the supermassive black hole at the Galactic center, appears to be 6 pc below the plane; actually, it is precisely at the center, but the IAU Galactic coordinate system assumes the Sun is exactly in the plane. There are, however, other estimates of the Sun's z-height that range from about 15 to 30 pc. These are mostly from optical or infrared observations and usually define a reference plane more locally to the Sun than our approach.