An interview with Richard Plambeck

Interview by Bo Reipurth, SFN #357 - September 2022

Your PhD from 1978 dealt with CO observations of interstellar molecular clouds. How did you choose this subject and who was your advisor?

My advisor was Ray Chiao, who worked closely with Charles Townes in the Berkeley Physics Department. New instrumentation for astronomy was a key focus of the Townes/Chiao group. I was particularly interested in applied solid-state physics, and Ray had lots of exciting ideas for sensitive new microwave detectors based on superconducting weak links and the Josephson effect. I worked on one of these ideas for awhile; the hope was that it would prove useful for observations of the 2-1 CO line that Paul Goldsmith and Townes were planning. However, progress was slow and it proved difficult to get repeatable results, so, with Ray's blessing, I began helping Paul with the more conventional Schottky diode receiver that he had developed.

Our first observations of the 2-1 line with this receiver were made at Lick Observatory using the 120-inch optical telescope during the day. Later, with help from Dave Williams and Tap Lum of the Berkeley Radio Astronomy Lab, we rebuilt the receiver to use it on a 6-m radio telescope at the Hat Creek Observatory. After Paul graduated, I continued the observations at Hat Creek, spending many weeks at the observatory waiting for rare periods of low humidity and tolerable atmospheric extinction. I felt lucky when the system temperature dropped below 6000 K. It's striking to think that now, with 230 GHz system temperatures of 100 K at ALMA, an observation that required many hours at Hat Creek can be done in a few seconds.

Anyway, the main result of these observations was that toward most sources, the 2-1 and 1-0 CO line profiles were strikingly similar. It seemed that the two transitions were probing the same parcels of gas, even though the 2-1 line was expected to have 2 to 3 times higher optical depth than the 1-0 line. This pointed to the prevalence of large-scale velocity gradients or clumpiness in molecular clouds.

You were part of the team that discovered the bipolar outflow in L1551, and in 1983 you followed this up by a larger study of clumpy outflows in star forming regions.

Yes, I was fortunate that Paul VandenBout invited me to bring the 230~GHz receiver to the Millimeter Wave Observatory in West Texas for an observing run with Ron Snell and Bob Loren. Probably not many people know about the MWO (which closed in 1988), but it had a high precision 4.9-m telescope with a gold-plated surface and an Invar backup structure for thermal stability. The MWO dome was located downhill from the big optical telescopes at McDonald Observatory; radio observers slept in a wind-buffeted trailer just across the road from the control room. I loved the place. There was a great view of the surrounding mountains, and the skies were impressively dark -- in late winter, for many hours after sunset, the zodiacal light was visible as a magnificent cone extending high above the western horizon. And, because the site was higher and drier than Hat Creek, observations were much faster.

An early focus of our MWO 2-1 CO observations was self-reversed line profiles to find evidence for cloud collapse. However, in a 1-0 CO map of L1551, Ron and Bob discovered a bipolar pattern of redshifted and blueshifted line wings centered on infrared source IRS-5. This structure was roughly 30 arcmin long, about the diameter of the Moon, so it was apparent only in the very large map that they'd made. We decided to observe one of the L1551 positions in 2-1 CO. The 2-1/1-0 line ratio was as high as 3 in the line wing. Finally! After observing so many sources where the 2-1 and 1-0 line profiles were boringly similar, it was exhilarating to find a source where $^{12}$CO was unambiguously optically thin. From the 2-1/1-0 line ratios, we were able to estimate the temperature and mass of gas in the outflow shell, and to show that this shell was plausibly material swept out of the two lobes by a stellar wind.

It turned out that the large 2-1/1-0 line ratio we observed toward L1551 was an anomaly. In the next few years we made 2-1 CO observations of more outflows at the MWO, using an improved receiver constructed by Neal Erickson at UMass, and compared these with matched beamwidth 1-0 CO observations from Kitt Peak. In most sources both CO transitions were optically thick even in the weak high velocity line wings. Since this high velocity gas typically was extended relative to the telescope beam, we argued that the emission arose from numerous dense clumps within the outflows.

In 1987-88 you and your collaborators published a series of papers on the Kleinmann-Low nebula. What did you learn?

Following the discovery of the bipolar outflow from L1551, I was immediately convinced -- incorrectly, as it turns out -- that the broad CO line wings, or ``plateau'' source, in Orion would prove to originate in a bipolar outflow driven by a massive star near the infrared source IRc2. By this time, I had begun working as a research astronomer in the Berkeley Radio Astronomy Lab, constructing 3mm receivers for the Hat Creek Interferometer. We set out to image the Orion outflow using this (1-baseline) interferometer.

Initially, it was possible to observe only over a narrow frequency range near 86~GHz, so we imaged the region in SO rather than CO. Frustratingly, our map (with about 6 arcsec resolution) showed that the SO emission was extended in a NE-SW direction, perpendicular to the 2 lobes of vibrationally excited H₂ that presumably marked the outflow. This led us to conclude that the SO line originated in a dense, expanding torus of material in the equatorial plane of the outflow, a hypothesis that seemed to be confirmed when Neal Erickson and colleagues mapped the 3-2 transition of CO and found a vaguely bipolar structure aligned with the H₂ lobes, although the apparent origin of this outflow was surprisingly far -- about 10 arcsec -- from IRc2.

In the next few years we added a third telescope at Hat Creek, and one of our graduate students, John Carlstrom, developed a local oscillator that was tunable from 68 to 115 GHz, allowing observations of spectral lines anywhere in the 3mm window. These days multiplier chains make it easy to observe any frequency you want, but back in 1985 John's Gunn oscillator was a giant leap forward -- no more expensive klystrons, no more high voltage klystron power supplies. John later started a business on the side that supplied these oscillators to observatories around the world. It was one of Berkeley's most significant contributions to radio astronomy instrumentation.

Taking advantage of the newly expanded tuning range, we mapped a wide variety of molecules in Orion over the next few years -- HCN, HCO⁺, HDO, CH₃OH, CH₃CN, and so on. We located the hot core and compact ridge sources that long had been identified in single dish line profiles, and tried to measure temperature gradients across the hot core clump to determine whether it was self-luminous or was heated by IRc2 -- although, with angular resolutions of a few arcseconds, it was nearly impossible to say.

With a 3-baseline interferometer it required many months to move the telescopes through enough configurations to accumulate the data for each map. Mapping speed improved dramatically after the University of Illinois and University of Maryland joined Berkeley to create the BIMA consortium, and the array was expanded to 6, and later 10, telescopes.

We then made a renewed push to map as many species as possible in Orion. I had the naive idea that we would first identify all the clumps in the region, then generate a table of chemical abundances in each clump. So we mapped the first molecule, say HCN, found the centers of all the clumps, then went on to the next molecule, only to find that the clumps now were in slightly different places. This pattern continued, molecule after molecule, until I was sorely tempted just to average together all the maps and label the result “C0” (C-zero). Fortunately Mel Wright and I persuaded Dave Wilner, then an astronomy graduate student, to help us summarize the results in a more coherent fashion.

Masers have been a long term interest for you, among other regions you and Karl Menten analyzed methanol masers in the DR 21 region in a highly cited study from 1990. What were some of your results?

If you build an interferometric array, masers will become your close friends. They're point sources, visible on all baselines, and often enable you to self-calibrate the data. Measuring their positions, velocities, and flux densities is fun. In fact, the very first scientific result we obtained at 3mm with the single baseline Hat Creek Interferometer was to show that the SiO maser in Orion was located within 1 arcsec of IRc2.

Class I methanol masers are fascinating because they typically are far offset from infrared sources, compact HII regions, or other masers. In the 1990 DR21 observations we mapped the 95 GHz methanol masers and 98 GHz CS line simultaneously, and found that the masers tended to be clustered along the edges of CS clumps. This hinted that the masers formed where dense clumps were shocked by outflows, heating the gas and enhancing the methanol abundance.

More recently you and your collaborators have published several papers on Orion source I. What is the picture that is emerging of this source?

Ah, yes, Source I. (I've begun abbreviating it SrcI). SrcI is the massive young star, or binary, about 1 arcsec from IRc2 in the heart of the Kleinmann-Low Nebula. SrcI is so deeply embedded in dust that it's directly observable only at radio wavelengths. It's the dominant source of luminosity in the region -- IRc2 turns out to be just a nebulosity that is heated by SrcI. Since Orion-KL is only 400 pc away, you'd think we'd understand SrcI very well by now, but I'm not sure that is the case. Long ago I was fond of proclaiming that “it's hard to be 100% wrong in astronomy,” but SrcI has taught me otherwise. In fact, not only is it possible to be 100% wrong, it's possible to be 200% or 300% wrong, by coming up with model after model that later are disproved. I'll give you some examples.

An early mistake was to assume, as I mentioned earlier, that SrcI (presumed to be IRc2 in the early days) drove the high velocity outflow in Orion. Although we never found convincing evidence of this in interferometric CO images, observations of the SiO masers associated with SrcI seemed to support the idea. For example, in some of the early observations at Hat Creek, Mel Wright and I found that the maser's two main velocity components at -5 and +16 km/s were spatially offset from one another by 0.14 arcsec along the NW-SE axis of the putative high-velocity outflow. This separation was tiny compared to the fringe spacing provided by the interferometer, so we worried that it might be an instrumental effect caused by phase ripple in the passband. This prompted me to set up an 86~GHz transmitter at a fire lookout tower above the Hat Creek valley. We observed the transmitter as it swept back and forth across the maser frequency, allowing very accurate measurements of the passband. Alas, the transmitter's frequency drifted with temperature, forcing me to drive up the gravel road to the lookout tower at 3~a.m. to tweak it back into range, but the measurements confirmed that the passband was flat and that the offset of the 2 maser features was significant.

A few years later, after expanding the array to 3 telescopes, we were able to detect even more structure in the maser -- it appeared that the emission originated from 2 arcs, with a smooth velocity gradient along each arc. Richard Barvainis had written a beautiful paper proposing that the linear polarization pattern measured for the Orion SiO maser could be explained if the masers originated in an expanding, rotating disk. I was excited to find that our position data also fit an expanding, rotating disk model, with the plane of the disk perpendicular to the high-velocity outflow. Everything fit neatly together.

Today we know that this appealing picture was completely incorrect. SiO maser emission does indeed originate in an outflow from SrcI, but the outflow is along a NE-SW axis, perpendicular to the one we'd assumed. The offset of the two main velocity components is real, caused by rotation at the base of the outflow, but the smooth arcs were artifacts caused by fitting centroids to clusters of maser spots; VLBI observations show that the masers actually are grouped into 4 main clumps. Finally, we now recognize that the “bowtie” structure that was visible in BIMA maps of the v=0 SiO line (a mixture of thermal and maser emission), and that we'd interpreted as a dense disk, is actually the bipolar outflow itself, a fact that became obvious when we obtained improved images of the source with CARMA.

Another of my mistakes had to do with proper motions. By 1995 we'd extended BIMA's longest baselines to 1.3 km, providing subarsecond resolution in the 3mm band. This allowed us to measure the separation between SrcI and the Becklin- Neugebauer Object (but not their absolute positions) to an accuracy of better than 0.02 arcsec. The result disagreed with the separation measured at the VLA several years earlier, so we concluded, correctly, that one or both sources were moving. However, because SrcI was deeply embedded in molecular gas while BN was not, we argued that SrcI was stationary and that BN was a runaway B star. Later, Jonathan Tan used additional Hat Creek data to show that BN probably had been ejected from Ө¹ Ori C, a 45 M_⊙ binary in the Trapezium.

Once again, everything fit together neatly. Yes, it was odd that BN's proper motion had carried it almost exactly past SrcI, but perhaps this was a coincidence. Thus, I was dumbfounded when a careful analysis of VLA data by Luis Rodriguez and colleagues found that both SrcI and BN were moving and that they were recoiling from one another, probably following the decay of a multiple system. Shortly thereafter, Luis Zapata and collaborators showed that the high-velocity gas in Orion was not a classical bipolar outflow at all; rather, it consisted of ejecta from an explosive event associated with the decay of the multiple system.

A key prediction of the dynamical decay model, based on the recoil speeds of SrcI and BN, is that SrcI's mass should be 15-20 M_⊙. For a long time I argued that this was unlikely, based on fits to 0.2 arcsec resolution ALMA spectral line rotation curves that suggested a central mass of just 5-7 M_⊙. Admittedly, these estimates required fitting centroids to emission regions that were barely resolved by the synthesized beam. However, I was struck by the fact that a similar mass estimate of of 7-10 M_⊙ was obtained from velocities of SiO masers, which are very well resolved in VLBI observations. Once again, however, I was mistaken. More recent, higher angular resolution, observations of rotation curves of H₂O and NaCl lines toward SrcI by Adam Ginsburg demonstrate unmistakably that the central mass is 15 M_⊙. Apparently these lines are better tracers of material in the gravitationally supported circumstellar disk -- NaCl, in particular, seems to be destroyed in the outflow -- whereas SiO masers originate in outflowing gas which is not gravitationally bound to the central star.

Well, I've spent enough time discussing mistakes, not that this was a comprehensive list! Perhaps the message is that one should not become too attached to any model, no matter how attractive it may seem.

What are your current interests?

I'm a bit embarrassed to admit that Orion Source I still is one of my interests. It would be great to find conclusive evidence that the central object is a close binary, as predicted by the dynamical decay model, but the binary separation is expected to be only a few AU. The center is heavily obscured by dust at mm wavelengths, so cm observations probably will be necessary to peer into the interior. I'm sure there are still big surprises in store, but probably it will take the ngVLA to see them.