An interview with Harold Yorke

Interview by Bo Reipurth, SFN #343 - July 2021

What was your thesis about and how did your interest in star formation start?


At the time of my Masters and PhD theses (1971-1974) my advisor, Rudolf Kippenhahn (a stellar evolutionist), and his group at the Göttingen University began studying various aspects of ISM evolution in order to understand the events leading up to the Hayashi phase of stellar evolution. I joined his group as a Fulbright Scholar from CalTech, where I previously had worked on X-ray rocket missions with Gordon Garmire in preparation for my PhD. My interest in hot plasmas meshed well with Kippenhahn's new thrust, and he asked me to develop a computer code for ISM heating and cooling and use it to determine how the warm ISM cools down to neutral HI clouds. Working with Werner Tscharnuter, Peter Biermann, and Kippenhahn, we used my code to study this phase transition after passage through a spiral density wave shock. This work became my Master’s thesis and piqued my interest in the not-so-hot ISM. I remained in Germany as a Fulbright Fellow, leaving both Caltech and X-ray astronomy. My 1974 PhD thesis (Stösse zwischen interstellaren Wolken) was a hydrodynamic/thermodynamic study of the evolution of HI clouds undergoing collisions. The code included H2 molecule formation and destruction. During the collisions, the compressed parts of the clouds became fully molecular and grain growth was enhanced, resulting in depletion of the main coolants in the HI phase, e.g. carbon, nitrogen, and oxygen.


One of the justifications for the thesis was to determine whether star formation could be initiated during such collisions. Stability analysis of the thin, highly compressed sheet demonstrated that the growth rate of density perturbations was too slow. The compressed regions re-expanded before collapse could be initiated. Of course, the HI clouds in this study were pressure-bound within the warm ISM to begin with and far from gravitationally bound. Had I considered collisions between massive molecular clouds on the verge of gravitational collapse, the answer would certainly have been different.


One of your most cited papers is the 1977 study that you and Endrik Krügel did on the dynamical evolution of massive protostellar clouds. How has the field developed over the last 40 years?


Parallel to my Masters and PhD work, another Kippenhahn student, Endrik Krügel, developed a hydrodynamic code for studying the expansion of HII regions into a constant density neutral HI medium, including both H and He ionization. We finished our PhD theses at the same time and decided to pool our resources to continue working on related topics. It was clear to us that HII region expansion into a constant density medium was unrealistic; the process of producing a star massive enough to ionize its surroundings must surely mold the physical conditions in the star's immediate vicinity, long before the star emits its ionizing radiation. Thus, we considered how to produce such massive stars, resulting in the above-mentioned 1977 study. It was meant to be a prequel to a more comprehensive study of HII region evolution.


Although our work demonstrated the importance of radiative acceleration on dust grains during the formation of massive stars, we realized that our simplifying assumptions were inadequate to truly simulate the evolution of massive protostars. Great progress has since been made through 1) abandoning the assumption of spherical symmetry, 2) improved methods of radiation transfer, 3) detailed studies of grain properties and grain growth, 4) simultaneous stellar evolution of the accreting central (proto) star, 5) the star's feedback forces on the surrounding material, 6) the influence of magnetic fields, and 7) the launching and collimation of disk winds and outflows. I feel fortunate that I have been able to lead or participate in such studies and/or inspired others... not to mention, being inspired by the wealth of new data in the infrared and by the insights others have provided. Still, there is much more work to be done. No numerical simulation covers all of the above-mentioned aspects. Moreover, stars in general and massive stars, in particular, do not form in isolation. What is the effect of interactions between neighboring protostars in a forming cluster? Why is the multiplicity of massive stars so high (e.g. binaries, triplets, quadruple systems in the Orion Trapezium)? What about star formation in extreme environments, e.g. the center of the Galaxy and in Starburst galaxies?




Your Annual Reviews article from 1986 surveyed the dynamical evolution of HII regions. Looking at the subject then and today, are some of the issues still unsolved?


The formation and evolution of HII regions are intimately related to the formation and early evolution of massive stars. In the years leading up to that 1986 review, we had just begun to consider these two fields as a whole. Also, we realized that massive stars form in molecular clouds of a finite extent. When an ionization front reaches the molecular cloud boundary, the pressure difference between ionized cloud material and the cloud's exterior results in what we (Tenorio-Tagle, Bodenheimer, and I) called the "champagne flow". These were important new aspects of the understanding of HII region evolution. Not included in that 1987 review was the realization that even massive protostars form disks and moreover, these massive protostars are always in the vicinity of lower mass protostars with disks. These disks provide a huge repository of material for compact and ultra-compact HII regions and proplyds within HII regions via "photoevaporation". Our lack of a full understanding of the evolution of HII regions is related to our ignorance of the complexity of the local environment in star-forming regions.


In the 1990s you wrote, with Peter Bodenheimer and Greg Laughlin, a series of papers on hydrodynamical calculations of the formation of protostellar disks. What were the key new insights?


Dating back to the work of Kant and Laplace in the 18th century, disks have long been considered a key ingredient in the formation of our Solar System and likely other planetary systems. Prior to our 1990s studies, it was known that even a small amount of angular momentum in the pre-collapse clump leads to the formation of a flattened disk. However, whether or not a central hydrostatic core can form à la Larson (1977) depends either on an exceedingly small amount of initial angular momentum (Larson assumed zero in his initial study) or on sufficient angular momentum transport within the disk. In our 1990 studies, we focused on the formation and early growth of the disk and of the central core ("protostar") due to the transfer of angular momentum. Shortly after disk formation and during much of the early highly embedded phase of evolution, the disk mass is relatively high, between 1/2 and 1/3 of the protostellar mass; i.e. far more massive than the ''minimum mass Solar Nebula'' or even T Tauri disks. The spectral appearance of embedded protostars varies not only as a function of age but also as a function of viewing angle. For example, an edge-on view of the embedded disk + protostar could be mistaken for an earlier evolutionary phase. All of the above is true for both low mass and intermediate-mass protostars (and presumably also for massive protostars). Later studies confirmed this for the high mass case.



You have also been interested in the destruction of disks around massive stars. What is more calamitous for a disk, photoevaporation, or stellar winds?


You have touched on one of the basic themes of my combined interest in high mass star formation and HII regions. Disks are intimately involved in the star formation process and once a high mass protostar reaches the main sequence, it develops a wind accelerated by UV radiation. That UV radiation also begins to ionize the star's surroundings. Initially, "surroundings" include the remnant circumstellar disk. In numerical studies without stellar winds, the stellar UV radiation ionizes and heats material at the surface of the disk, causing it to expand and ultimately flow away from the disk ("photoevaporation"). The timescale for this process to destroy the disk is roughly 105 years. Simultaneously with photoevaporation, the stellar wind creates a local bubble of hot material around the star, pushing the inner disk boundary outward. Much of the pent-up energy of this hot bubble is released upward and downward along the rotation axis, i.e. along the path of least resistance. The net result of including the effect of stellar winds is a slightly shorter timescale for disk destruction. Because these powerful stellar winds are UV radiation-driven, stellar winds from hot stars and photoevaporation always occur together. I am unaware of numerical studies of the effects of external stellar winds on a protostellar disk, but I would still argue that you cannot have one without the other.



The subject of massive star formation has gained enormous interest in recent years, and your 2002 study with Cordula Sonnhalter and the more recent Annual Reviews article with Hans Zinnecker are heavily cited. Have we finally understood massive star formation?


Absolutely not. I hope that my previous answers have touched on some of the many things that we still do not understand well. In addition to all of the above, the study of Pop III star formation and star formation in extreme environments is still in its infancy. Are the feedback effects I mentioned above the reason for an upper mass limit and how does this change with metallicity?


You recently retired as director of the Stratospheric Observatory for Infrared Astronomy. What enticed you to take on that role, what do you consider to be your main accomplishments, and how do you see the future of SOFIA?


After leaving Germany and coming to JPL in 1998, I became increasingly interested in far-infrared space missions. Inspired by the TPF/SIM mission concepts I submitted an idea to NASA for a cryogenic far-infrared space interferometer, SPIRIT, with a 30m baseline. NASA provided funding for me to work with Dave Leisawitz and the industry to develop a more mature mission concept. Shortly thereafter, SPIRIT was included in and received an honorable mention in the 2000 Decadal Review. A few years later I developed a concept for an off-axis cryogenic 4x6m filled aperture far-infrared telescope, CALISTO. I was PI of a MIDEX proposal to NASA for the Far Infrared Line Mapper (FILM) that would have mapped the entire Milky Way plane and selected nearby galaxies in the C+ 157μm and N+ 205μm lines. FILM received a high technical rating but lost out to WISE. From 2001 to 2006 I was the NASA Project Scientist for Herschel during its development phase and provided scientific-technical oversight to NASA's contributions to that mission.


I mention this abridged version of my mission work to demonstrate that I was an enthusiastic supporter of far infrared missions. When the position of SOFIA director became available, I was well aware that SOFIA had had two prior near-death experiences: Twice the President's budget had zeroed out SOFIA funding, only later to be restored by Congress. Realizing that I was taking a risk, I nevertheless applied and accepted the subsequent offer. I considered SOFIA to be extremely important for far-infrared astronomy and star formation studies.


During my tenure as SOFIA Director of Science Mission Operations, it was imperative that SOFIA transition from a more-or-less "insider" facility to one of NASA's key astronomical assets, available to all astronomers. Of course, I exaggerate; SOFIA was always available to all astronomers. However, more work had to be done in terms of community outreach, offering in-person and online training courses, and improving the interface with users. Another issue at the time was the poor completion statistics of observing programs. Constrained by fuel and flight safety rules, SOFIA has to return to base after 8 hours of observing. Thus, a theoretical maximum of 4 hours of flying in one direction, staring at one object, is possible. In practice, flight observing "legs" are much shorter than 4 hours and a given flight serves several programs. When I became director in 2016, projects requiring several flights were too often left only partially complete, whenever technical "grounding" issues or extremely poor weather forced us to cancel flights. For a number of technical reasons it was difficult to swap flight plans from one day to the next, so lost flights could not easily be made up. Even if you could swap out flights, someone else or many others would lose part of their observing programs. In this sense, SOFIA was treated as a ground-based observatory: i.e. if you lose your observations due to bad weather or technical issues, well tough luck. Better luck next cycle...



The SOFIA aircraft has since become more reliable as NASA became better at operating this aging aircraft, either anticipating potential grounding issues or fixing them on a shorter time scale. We increased our flight cadence, we introduced a new prioritization scheme of accepted proposals and their distribution onto flights that improved the completion statistics of high and medium priority observations. Prior to this change, basing a PhD proposal on future SOFIA data could be a risky endeavor. We became more adept at creating improved, more efficient flight plans and making adjustments to and swapping flight plans on an increasingly shorter time scale. Lost flights of highly rated observations could be made up. Our data reduction pipelines matured further. With the help of IPAC, we improved the accessibility of archival data at IRSA. We introduced new funding opportunities: e.g. the archival research, thesis-enabling, and SOFIA Legacy programs.



As any director of an astronomical facility, I was interested in utilizing my Director's Discretionary Time to occasionally push the limits of what our instruments could accomplish. Observing the Moon with the short-wavelength grism of FORCAST was a prime example. In a brief one-hour observing leg in 2018, SOFIA detected unexpected large amounts of water on the sun-lit face of the Moon away from the poles. This DDT observation resulted in a 2020 Nature publication and was widely discussed in the general press and public newscasts. A new SOFIA Legacy Program was created to map the potential landing sites for NASA's upcoming VIPER Mission, a lunar rover that prospects for water.



It is hard to overemphasize the importance of SOFIA for the future of far-infrared astronomy. Not only is SOFIA the only far-infrared facility available to all astronomers, but it will also remain so for at least another decade. SOFIA continuously allows follow-up studies of important new results such as the discovery of water in the Clavius Crater, as opposed to Herschel, for which the observation schedule was carved in stone well in advance and could not easily react to new results or insights. Unlike for space missions, SOFIA's instrumentation can be upgraded, and -- depending on available funding -- new state-of-the-art instruments can be built. Because we can make repairs and adjustments to the instruments between observing campaigns, the technical readiness level of SOFIA's instruments is not restricted to the high level demanded for space missions; it is a well-known fact that space instrumentation is no longer state-of-the-art by the time of launch. Continued improvement of far infrared instrumentation is vitally important in retaining and expanding current expertise. Unlike the case for the X-ray, optical, or near-infrared regimes, we cannot rely on industry or the military to further improve our detectors. We also cannot afford to lose the expertise of using these detectors, waiting for the next big mission.



In the President's FY 2022 budget request, SOFIA is again slated for termination. In the two previous attempts to cancel SOFIA, Congress has weighed in and restored its budget to workable levels. I sincerely hope that Congress remains convinced that the overall costs of SOFIA are justifiable in the long term. Until our next space-based far-infrared observatory comes online, SOFIA fills an important gap between the radio and near-/mid-infrared spectral regimes. Observational studies of early phases of star formation would be severely hampered without it.