Speed check on the ‘fastest’ star in Gaia

Title: Lessons from the curious case of the ‘fastest’ star in Gaia DR2

Authors: D. Boubert, J. Strader, D. Aguado, G. Seabroke, S. E. Koposov, J. L. Sanders, S. Swihart, L. Chomiuk, and N. W. Evans

First Author’s Institution: Magdalen College, University of Oxford, UK

Status: Accepted in MNRAS, closed access

‘Hypervelocity’ stars move fast. So fast they can escape from the gravitational pull of the Milky Way. A few tens of hypervelocity stars are known but the Gaia space telescope is expected to find hundreds more as it maps the positions and velocities of stars in the sky. A star’s velocity is split into its tangential velocity (how fast it moves across the sky, a.k.a proper motion) and radial velocity (how fast it moves towards or away from us). Combining them together using trigonometry gives the star’s total velocity.

Measuring radial velocities

As the Gaia pipelines are still being improved to combine these velocities, only stars with radial velocity greater than the escape velocity of the Milky Way (600 km s-1) are guaranteed hypervelocity stars. Currently one star meets this criteria: Gaia DR2 5932173855446728064 (which we’ll call Gaia ‘064 for short). Gaia ‘064 has an incredible radial velocity of -614 +-2 km s-1, calculated as the median of seven individual measurements. Figure 1 shows Gaia ‘064 is in a very crowded field on the sky. It has nine neighbouring stars within 8 arcsec, far more than other hypervelocity stars tested.

Figure 1: 100 arcsec view of Gaia ‘064 (left) and 20 arcsec zoom in with stars identified in Gaia marked by green circles (right). Figure 1 in today’s paper .

As Gaia ‘064 is in an abnormally crowded field, the authors took follow up spectra to verify the Gaia DR2 radial velocity. Using eight spectra from the SOAR telescope taken between 5th May and 16th September 2018, they calculated a median radial velocity of -56.5 +-5.3 km s-1.

Explaining the difference

Clearly, the Gaia radial velocity is significantly larger than the follow up value. The authors explore two explanations, one where the Gaia measurement is true and another where it is incorrect.

Scenario 1: Gaia was right

Stellar radial velocities can change by several hundred km s-1 in binary systems as the star orbits its companion. The measured radial velocity is then the sum of the star’s orbital velocity and how fast the system is moving towards or away from us. However, testing this scenario is hard as Gaia does not record the times it took the radial velocity measurements. We only have 42 predicted times over two years when Gaia could observe the target from the ESA Observation Forecast Tool. The radial velocity cannot have changed by ~500 km s-1 between individual measurements, otherwise Gaia ‘064’s median error would be much higher than 5 km s-1. However 26 of the possible 42 times occur within 3.75 days, so all seven measurements could be made then and still be consistent with each other if the binary is long period.

Combining all 15 radial velocity measurements, the authors find the orbital period must be longer than 1000 days. They find the minimum mass of the unseen companion is 3000 solar masses, making it an intermediate mass black hole (IMBH). Gaia ‘064 does have an unusually large number of neighbouring stars, which is expected around an IMBH. However, Gaia ‘064 is young, whereas IMBH are expected to host much older stars. Additionally, neither of two neighbouring stars the authors examined had extreme radial velocities, suggesting that Gaia ‘064 is not orbiting a black hole.

Using just their eight follow up radial velocities, the authors find the target is more likely in a close stellar mass binary with a radial velocity of -62.5 +- 5 km s-1, than on its own or in a long period binary. This is not conclusive as the data are taken close together in time, so the target could still be in a long period binary.

Scenario 2: Gaia was wrong

Quality cuts have already been applied to remove unreliable radial velocities, so perhaps Gaia ‘064’s measurement is spurious due to its unusually high number of close neighbours. Gaia’s spectrograph records the spectra continuously as it scans the sky. The pipeline can separate measurements for stars more than 6.4 arcsec apart. If the scanning direction (shown by orange lines in Figure 2) happened to pass through two stars closer together than 6.4 arcsec, their light blends in a single spectrum.

Figure 2: 20 arcsecond view around Gaia ‘064. The 26 scans of Gaia ‘064 which also pass through Gaia ‘352 are shown as orange lines. Figure 7 in today’s paper.

Most of the time, blended measurements should be filtered out as two sets of lines implies a spectroscopic binary. Gaia’s spectrograph will offset blended light by 145.1 km s-1 per arcsecond. One close neighbour, Gaia ‘352, is only 4.28 arcsec away, giving an offset of 619-624 km s-1. Subtracting this offset from Gaia ‘352’s radial velocity of 5 km s-1 gives -613 to -618 km s-1, consistent with the Gaia ‘064’s Gaia radial velocity of -614 +- 2 km s-1. Therefore, the anomalous median radial velocity can be explained if the spectra were blended in all seven measurements. Combined with the small error in the radial velocity measurement makes it likely that all seven measurements occurred in that 3.75 day window and passed through both bright stars (shown in Figure 2). Therefore the light is likely blended and Gaia ‘064 an average, non-hypervelocity star.

Cleaning the catalogue

Many thousands of stars with Gaia radial velocity measurements have another source within 6.4 arcsec that also has a radial velocity measurement or is as bright. The Gaia colours pipeline discards the colours of stars where blends were suspected, so if a star has a radial velocity measurement but is missing colour measurements then it may be a blend. To obtain a much cleaner catalogue, at the expense of completeness, the authors propose to remove stars which:

  • Are missing either colour magnitude.
  • Have fewer than four radial velocity measurements
  • Have a neighbour within 6.4 arcsec which has a radial velocity measurement or is brighter.

Applying the above quality cuts to the 202 stars with radial velocities greater than 500 km s-1, only 90 stars survive. While some real hypervelocity stars are likely removed, this leaves a much cleaner sample for investigating hypervelocity stars.

Mapping hypervelocity stars

Hypervelocity stars may reveal something about the systems they are shot out of, including massive black holes and supernova, and about galactic dynamics more generally. Unfortunately it seems that Gaia ‘064 is not part of this group. The case of Gaia ‘064 is a cautionary tale that if a number of unlikely scenarios line up, the results can be very wrong, so it is always worth checking!


Featured image: Artist’s impression of Gaia with the Milky Way in the background. Copyright: ESA/ATG medialab; background image: ESO/S. Brunier.

About Emma Foxell

I am a PhD student at the University of Warwick. My project involves searching for transiting exoplanets around bright stars using telescopes on the ground. Outside of astronomy, I enjoy rock climbing and hiking.

Leave a Reply