Massive Stars

Dunlap Postdoctoral Fellow, Dunlap Institute at the University of Toronto

2021 – present

I recently began a postdoc position at the Dunlap Institute where I’ll continue studying evolved massive stars. Stay tuned as I make new discoveries!

University of Washington

2017 – 2021

I completed my PhD at the University of Washington under the supervision of Emily Levesque in 2021. My dissertation focused on the binary fraction of Red Supergiants in the Local Group Galaxies M31, M33, and the Magellanic Clouds. Evolutionary models show that binary RSGs will always exist in systems with B-type stars and thus spectroscopically they will show the strong TiO bands characteristic of RSGs as well as the upper balmer lines found in B-type stars. Combining the spectra of these two types of stars creates a “flatter” spectral energy distribution than expected and thus they can be picked out photometrically. By placing color-color cuts on archival photometry, I came up with a list of candidate RSG+B star binaries. The candidates were then observed spectroscopically and I ended up finding ~250 of them! I additionally determined the binary fraction across different environments and determined that the percentage ranged from 15-40%.

Lowell Observatory

2009 – present

I first started working with Phil Massey as his REU student in the summer of 2009. Even though the project he picked out for me was a ridiculous choice for a summer student (learn more about it below), I decided to continue working with him after returning to Wellesley in the fall. This collaboration has continued as I’ve switched back and forth between a part and full time employee at Lowell while both living in Flagstaff and working remotely. During these past 9 years, Phil and I have done a lot of great research on a variety of interesting projects.

Surveying M31, M33 and the Magellanic Clouds for Wolf-Rayet stars

Wolf-Rayet (WR) stars are evolved massive stars, and the relative number of carbon-rich (WC-type) stars and nitrogen-rich (WN-type) WRs should vary with the metallicity of the host galaxy, providing a sensitive test of stellar evolutionary theory. However, past WR surveys have been biased toward detecting WC stars, as their emission-line signatures are much stronger than those of WNs. Using an interference filter technique, we surveyed M31, M33 and the Magellanic Clouds for Wolf-Rayet stars. As expected, we’ve found quite a few new WNs!

Characterizing a new type of Wolf-Rayet star: WN3/O3s

As part of our Wolf-Rayet survey in the Magellanic Cloud, we stumbled upon a new type of Wolf-Rayet star! We’ve now found 9 of them in the Large Magellanic Cloud and we’ve modeled their spectra using CMFGEN. We’ve convinced ourselves that these new stars aren’t binaries (for a variety of reasons), and we’ve found that we can model their spectra using a single set of physical parameters. We’re still trying to figure out the evolutionary properties of these stars but the current theory has them being a short-lived phase of normal single star evolution at some metallicities – but stay tuned!

Examining the binary frequency of Wolf-Rayets in M31 and M33

Massive star evolutionary models generally predict the correct ratio of WC-type and WN-type Wolf-Rayet stars at low metallicities, but underestimate the ratio at higher (solar and above) metallicities. One possible explanation for this failure is perhaps single-star models are not sufficient and Roche-lobe overflow in close binaries is necessary to produce the “extra” WC stars at higher metallicities. However, this would require the frequency of close massive binaries to be metallicity dependent. As part of this research, we tested this hypothesis by searching for close Wolf-Rayet binaries in the high metallicity environments of M31 and the center of M33 as well as in the lower metallicity environments of the middle and outer regions of M33. After identifying ~100 Wolf-Rayet binaries based on radial velocity variations, we concluded that the close binary frequency of Wolf-Rayets is not metallicity dependent and thus other factors must be responsible for the overabundance of WC stars at high metallicities. However, our initial identifications and observations of these close binaries have already been put to good use as we are currently observing additional epochs for eventual orbit and mass determinations.

Measuring the masses of O-type binaries

Our search for O-type binary systems in the Magellanic Clouds began by taking a lot of pictures of regions rich in massive stars. After searching for stars that varied in brightness over time, we identified 48 periodic variables. We then spectroscopically confirmed these stars as binaries by identifying variations within their spectra that showed that the systems were moving. By measuring the movement of the stars, we were able to get orbit solutions and then very precise masses of the individual stars within the system! We then compared our results to evolutionary models.

Testing stellar evolutionary theory using yellow and red supergiants in the Magellanic Clouds

The yellow supergiant content of nearby galaxies provides a critical test of massive star evolutionary theory. By determining the number of yellow supergiants in a galaxy, we can compare our results with the Geneva evolutionary tracks. We originally completed this comparison using the yellow supergiant content of the Small Magellanic Cloud and found that the models over-predicted the yellow supergiant lifetime by a factor of 10. However, our friends in Geneva went back to the drawing board and made a few of their approximations a little less approximate. When we then did the same test using yellow supergiants and red supergiants in the Large Magellanic Cloud, we found that the new Geneva evolutionary models do an exemplary job at predicting both the locations and the lifetimes of these transitory objects.

Comparing two spectral modeling programs: CMFGEN and FASTWIND

Generally REU advisors chose projects that can be completed by the summer student within the summer (so 3 months). My REU project instead took 3 years to complete. Over those three years, I compared the results of two spectral modeling programs (CMFGEN and FASTWIND) by modeling O-type stars in the Magellanic Clouds. The two codes treat the modeling process quite differently. FASTWIND takes an approximate approach to dealing with all of the various elements while CMFGEN instead solves for each nitty gritty detail. Thus, while FASTWIND takes ~8 minutes to run, CMFGEN can take up to 48 hours. Both codes have been used extensively to model the spectra of O-type stars, but no comparison had been done using the different codes on the same spectra. After completing the modeling, we found that there was no significant difference in the temperature determinations, but the surface gravities did differ systematically. More information can be found in our “Bake Off” paper.

Sky Brightness at Kitt Peak National Observatory

Lowell Observatory

2009 – 2010

One of my research projects with Phil Massey involved measuring the night sky brightness at Kitt Peak National Observatory near Tucson, AZ. We then compared our 2010 data to measurements taken by Phil in 1990 and 2000. Remarkably, we found that the brightness of the Kitt Peak night sky at zenith is just as dark as it was 20 years ago. This suggests that the lighting ordinances established in Tucson and Pima County in the early 2000s have been incredibly effective. Our efforts were highlighted by the Tucson newspaper as well as numerous other sources such as Sky & Telescope and a press release by NOAO.


Wellesley College


During my junior and senior years at Wellesley I worked with Professor Richard French analyzing Saturn Cassini data. I spent the first semester investigating the azimuthal asymmetries in Saturn’s A and B rings using VIMS data. These asymmetries are caused by ring particle formations called wakes. Wakes are created by the inherent desire of particles to bind together through gravitational forces. However, these particles are within the Roche limit, meaning they can’t bond together and form a moon because of tidal disruptions from Saturn’s gravitational pull. However, the tidal disruptions do cause them to form small streams called wakes. These streams are tilted 20 degrees with respect to Saturn’s ring plane. Thus, depending on the tilt of Saturn’s rings and your viewing angle, it is either easier to see through the rings and the rings appear darker (trailing quadrant) or it is harder to see through the rings and thus the rings reflect more light back to you and appear lighter (leading quadrant).

The right image shows the rings in the left image reprojected. Thus, the brightening (in the leading quadrant) and darkening (in the trailing quadrant) is clearly visible in the A ring. (French et al., 2007)
Notice the brightening (in the leading quadrant) and darkening (in the trailing quadrant). (French et al., 2007)

I then spent the second semester looking at ring occultation data and measuring the location of different prominent ring features (such as the Cassini and Encke divisions). This data can then be used to help determine the precise pole solution for Saturn. Here is a picture drawn by a certain “clever” professor on my computer monitor while I took a break from measuring the locations of various ring features.



Wellesley College

2007 – 2008

As a sophomore at Wellesley I participated in the Sophomore Early Research Program. This research paired me with astronomy Professor Steve Slivan and his work on Koronis Family asteroids. During my first year of “real” astronomy research, I learned how to use the 24-inch telescope at Wellesley’s Whitin Observatory to collect and reduce CCD images. I then used these images to determine the periods and H magnitudes for three different Koronis Family members: (3032) Evans, (1443) Ruppina, and (2268) Szmytowna. The light curve for (2268) Szmytowna phased to the appropriate period is shown below.

SzmytownaFinalAt the end of the year, Professor Slivan and I published a paper in the Minor Planet Bulletin detailing our findings.