Vulcan II: The Wrath of Stellar Activity

Title: The Death of Vulcan: NEID Reveals That the Planet Candidate Orbiting HD 26965 Is Stellar Activity

Authors: Abigail Burrows, Samuel Halverson, Jared C. Siegel, Christian Gilbertson, Jacob Luhn, Jennifer Burt, Chad F. Bender, Arpita Roy, Ryan C. Terrien, Selma Vangstein, Suvrath Mahadevan, Jason T. Wright, Paul Robertson, Eric B. Ford, Guðmundur Stefánsson, Joe P. Ninan, Cullen H. Blake, Michael W. McElwain, Christian Schwab, Jinglin Zhao

First Author’s Institution: Department of Physics and Astronomy, Dartmouth College, Hanover, NH 03755, USA

Status: Published in The Astronomical Journal [open access]

It’s possible you may have heard of HD 26965, aka 40 Eridani A, aka the stellar host for the fictional planet Vulcan (homeworld of Spock, of Star Trek fame). Back in 2018, two teams of astronomers announced the likely presence of a super-Earth—Vulcan, perhaps—orbiting HD 26965, based on radial velocity (RV) measurements from a variety of instruments. These astronomers measured the mass to be equivalent to over 8 Earths, and the orbital period to be roughly 42 days. But before Trekkies could break out the Romulan ale, researchers noted that more work was needed to fully separate the RV signals from that of the exoplaneteer’s eternal enemy: stellar activity, which can mimic planetary signals.

The fate of Vulcan

Since then, study after study have highlighted stellar activity as the likely primary source of the RV signal, raising further uncertainty of the existence of Vulcan. What might bring us closer to a definite answer would be to combine detailed RV analysis techniques with what we already know about the effects of solar activity on radial velocity measurements. This knowledge can be applied to high-resolution spectra of HD 26965 taken by NEID at the Kitt Peak National Observatory. NEID represents the latest and greatest in ground-based spectrometers, ensuring clearer, more frequent data than previous studies had access to.

Important to determining the status of HD 26965 b (the standard name for our Vulcan) is the concept of phase lag. Based on what we know from studying solar activity, the observational parameters used to study stellar activity can become offset in time from the effects of stellar activity on RV measurements. This would mean that stellar activity might not have been fully corrected for in early studies of this planet. If a decaying starspot or plage were present on HD 26965, correcting for it would lead to a reduction in RV signal.

Combining RV analysis techniques

To account for this phase lag, the authors first compute what are essentially smoothened RV signals by taking the sum of every RV corresponding to every spectral line, weighted by their corresponding errors. These are corrected for NEID systematic quirks (such as by converting measurements into the stellar frame from the observatory frame NEID usually works in). This produces a “template” RV that the authors compare each observed spectral line to, noting differences between observed RV and “template” RV. This comparison allows them to view how phase lag may affect the amplitude and location of RV signal.

The hits just keep on coming

The observed NEID RVs have a similar 42-day period to the planet model proposed by an earlier study, but Figure 1 shows that they are out of phase by 30-40%! Straight off the bat, it’s not looking so good for Vulcan with strike number one. In addition, comparing these RVs to activity indicator lines Ca Ⅱ, H, and K, reveal that they share a similar period, but do not seem to be correlated. Strike number two!

A plot of two sets of data: one yellow, which represents the periodic planet model proposed by one early paper on HD 26965 b, and one purple, which represents the various NEID-observed radial velocities over the same period of time. The RVs take place at different points in time from the planet model, indicating a phase offset.
Figure 1: The NEID-observed RVs of HD 26965 (marked in purple) compared to the planet model proposed by an earlier study (Ma et al. 2018, marked in yellow). The NEID RVs are out of phase with the planet model by 30-40%. Source: Figure 5 from the paper.

Going back to phase lag, the authors calculate phase offsets for all activity indicators using a Gaussian process, finding a consistent phase lag between 4.65 and 6.67 days, over 10% of the star’s rotation period. Shifting data with this phase lag leads to significant reduction of RV signal strength when modeled with a 42-day period, and overall over a variety of periods, seen in the periodograms in Figure 2. Strike number three!

A series of plots that are divided into two left and right columns, and seven rows. Each row represents a different activity indicator. Each column on the left represents corrected RVs corresponding to each activity indicator as well as uncorrected RVs. On the right, each column shows periodograms - plots that show the probability of a given period being correct. The periodograms tend to be weakest around the supposed 42-day period proposed by early papers.
Figure 2: Left: A variety of corrected RV measurements per activity metric, marked with purple circles, accompanied by original RVs, marked with gray squares. Right: Periodograms for corrected RVs, marked in solid purple, and for original RVs, marked in gray dashes. The red diamonds denote periods with the highest remaining power. At the supposed 42-day period, the probability of that being the actual period (called the power) drops significantly for most activity indicators. Source: Figure 8 from this paper.

Stop, stop, he’s already dead!

These are just a small sampling of demonstrations in the paper—ultimately, they all seem to point towards stellar activity probably being the source of RV signal from HD 26965, unfortunately for Vulcan.

Fortunately for us, this paper aims to act as a Swiss army knife of sorts. It accomplished several things—first, the authors showed that phase lags may be important in determining the relationship between RV and whatever may have affected RV. Second, a bundling of analyses makes for a stronger case compared to a single analysis technique on its own, giving more solid credence to the stellar activity hypothesis.

Last but not least, the methodologies used in this paper can be used on other RVs from other stars, including our own Sun! In addition to testing this multi-pronged analysis on other types of stars, the authors hope to use it on our Sun during especially turbulent activity periods. By refining our understanding of how our Sun and other stars fluctuate through time, we better our exoplanet detection techniques, and our chances at finding habitable worlds.

Astrobite edited by Dee Dunne

Featured image credit: NASA. This image was modified to appear in black and white by the Astrobite author. 

About Diana Solano-Oropeza

I'm a first year astronomy PhD student at Cornell University, where I study exoplanets, stars, and habitability using Gaia data. I earned my B.S. in physics at Drexel University before entering the Bridge to the PhD in STEM program at Columbia University. There, I researched TESS-detected exoplanets for two years. My hobbies include practicing Muay Thai, fictionwriting, and playing video games. You can check out my website at

Leave a Reply