Revisiting hypervelocity stars after Gaia DR2

Introduction

The number of candidate hypervelocity stars has ballooned in the years since the discovery of HVS1 and today there are more than 500 candidates in the literature https://faststars.space (see Fig. 4). There is, however, reason to be skeptical of many of these candidates. While Brown et al. 2006; Zheng et al. 2014; Brown et al. 2014; Huang et al. 2017 have discovered a further two dozen hypervelocity candidates that are likely late B-type stars far out in the halo and with an extreme radial velocity, most of the candidates are late-type, high proper motion stars. In a majority of cases, the radial velocity is itself unremarkable and the ‘hypervelocity’ classification is driven entirely by a large proper motion measurement. However, as noted in Ziegerer et al. 2015, there is reason to be cautious. The authors assessed the candidates in Palladino et al. 2014. They were unable to confirm them, with the ground-based proper motions fingered as the likely culprit.

The origin of hypervelocity stars remains an intriguing and open question. The tidal disruption of binary stars by the supermassive black hole at the Galactic centre, leading to the ejection of one of the stars (Hills 1988), is considered the most likely possibility (Ginsburg & Loeb 2007; Brown 2015, e.g.) . However, there remains the possibility that the hypervelocity stars were ejected from elsewhere in the Milky Way’s disc and are either supernova runaways or were dynamically ejected from star clusters. Recently Boubert & Evans 2016 and Boubert et al. 2017 argued that the hypervelocity stars could possibly originate in the Large Magellanic Cloud. The early-type hypervelocity stars are found in the halo, while the late-type hypervelocity stars are found within several kiloparsecs of the Sun. Thus, these two populations probe different kinematic regimes and can potentially be used to distinguish between the formation scenarios. The question of whether there are any late-type hypervelocity stars lies at the centre of the hypervelocity star mystery.

The European Space Agency’s Gaia space telescope was launched in 2013 and on the 25th April 2018 delivered its second date release (Gaia Collaboration et al. 2016; Gaia Collaboration et al. 2018, Gaia DR2,) containing astrometry and photometry for 1,692,919,135 sources, based on the first 22 months of operation. This catalogue includes parallaxes and proper motions for an unprecendented 1,331,909,727 sources, typically with sub-milliarcsecond precision https://www.cosmos.esa.int/web/gaia/data. Gaia can thus revolutionise the study of late-type hypervelocity stars. It will allow accurate tangential velocities to be obtained for all extant late-type hypervelocity candidates.

The objective of this paper is to provide a comprehensive update on the status of the hypervelocity candidates in the literature after Gaia DR2. We specifically focus on the nearby, late-type candidates because these are the stars whose status is most likely to change with improved astrometry. In Section 2, we briefly cover the history of searches for late-type hypervelocity stars. Section 3 provides an overview of the landscape of hypervelocity star candidates and looks in detail at the one confirmed late-type hypervelocity star. In the Conclusions, we discuss the implications of our results. In the Appendix we present the Open Fast Stars Catalog whose creation enabled this work.

History of searches for late-type hypervelocity stars

Prior to Gaia DR2, a number of late-type hypervelocity candidates had been claimed in the literature. We define late-type as stars whose spectral type is F, G, K or M, including both dwarf and giant stars. Many of these identifications were based on cross-matches between spectroscopic surveys such as SEGUE (Yanny et al. 2009) and LAMOST (Cui et al. 2012) together with the SDSS-USNO proper motion catalogues (Munn et al. 2004; Munn et al. 2008). With the addition of photometric parallaxes, this gives the full space motion of the candidate. The orbit is integrated in a Galactic model to assess whether it is unbound. The radial velocity is usually secure, but photometric parallaxes typically have errors of $\sim 15$ per cent. Even the most carefully constructed ground-based proper motion catalogues tend to have some erroneous measurements, especially in the high proper motion regime.

As an example, Li et al. 2012 searched through Sloan Digital Sky Survey (SDSS) Data Release 7 and identified 13 F-type hypervelocity star candidates. They used SEGUE spectroscopy and proper motions from the SDSS-USNO (Munn et al. 2004). They argued from orbit integrations that 9 candidates emanated from the Galactic Center of disk, whilst the remaining 4 had a more exotic origin, such as tidal disruption of dwarf galaxies (Abadi et al. 2009). Palladino et al. 2014 also carried out a search in the SEGUE G and K dwarfs sample, again based on proper motions from SDSS+USNO-B (Munn et al. 2004). The fate of these candidates illustrates the pitfalls of such work. Many of the candidates were contested either because they are high velocity halo stars and therefore bound or because the ground-based proper motions are inflated (Ziegerer et al. 2015).

The LAMOST survey also proved to be a happy hunting ground for late-type hypervelocity star candidates. Li et al. 2015 claimed 19 low mass F, G and K type hypervelocity star candidates from over one million stars found in the first data release of the LAMOST regular survey. They combined LAMOST spectroscopy with SDSS-USNO-B (Munn et al. 2008) proper motions. Their final cleaned candidate list used only stars with reliable proper motions, high quality spectra and trustworthy astrophysical parameters. The candidates had probabilities of being unbound, as judged from Monte Carlo simulations of orbit integrations, in excess of 50 per cent. However, there were 8 high quality candidates with a probability in excess of 80 per cent.

We are not the first to realise the potential of Gaia as a purger of late-type hypervelocity candidates. Marchetti et al. 2017 trained a neural network to identify hypervelocity star candidates in Gaia DR1 and noticed that one of their candidates HD 5223 had previously been suggested by Pereira et al. 2012. The Gaia parallax indicated that it was much closer than previously thought. Given the history of the subject, Gaia Data Release 2 proper motions might well be expected to winnow the late-type hypervelocity candidates.

Results

We note that the distance distribution shown in Fig. 4 is biased by the way that the early-type hypervelocity stars were discovered. The Hypervelocity Star Survey (Brown et al. 2005; Brown et al. 2007; Brown et al. 2014) selected for blue, faint objects at high latitudes, because a B type star would require a large velocity to reach the halo within its lifetime. Thus by construction our sample of hypervelocity stars is biased towards stars at great distances. The existence or non-existence of early-type hypervelocity stars closer to the Galaxy will allow us to tell whether the hypervelocity stars have a Galactic or extragalactic origin.

2 Type Ia supernova donors and survivors

Both US708 (Hirsch et al. 2005) and GD 492 (Vennes et al. 2017) are thought to be associated with Supernova Ia. We discuss each briefly. Note that the other white dwarf hypervelocity candidate SDSSJ124043.01+671034.68 (Kepler et al. 2016) is confirmed with Gaia DR2 to be bound to the Galaxy.

Hirsch et al. 2005 initially conjectured that US708, a helium subdwarf O star, was formed in the merger of two helium white dwarfs during an interaction with the SMBH at the Galactic centre. However, Justham et al. 2009 proposed that this star was more consistent with having been the low-mass helium donor to a massive white dwarf and thence having being ejected by the resulting thermonuclear supernova Ia, and a subsequent spectroscopic and kinematic analysis confirmed this as the likely origin channel (Geier et al. 2015).

3 The remaining late-type hypervelocity candidate

Conclusions

In this paper, we have combined the historical data on high-velocity stars with data from Gaia’s second data release. There is a single candidate late-type high-velocity object (LAMOST J115209.12+120258.0) that has a reasonably high probability of being unbound from the Milky Way and thus hypervelocity. However, the overwhelming majority of the historical late-type high-velocity candidates are almost certainly bound to the Milky Way. This is a clear demonstration of the superiority of space-based astrometry from Gaia over the earlier ground-based proper motion catalogues. It is anticipated that further Gaia DR2 studies will reveal many late-type high-velocity candidates, which will be added to the Open Fast Stars Catalog when they are announced.

Acknowledgements

We thank Warren R. Brown, Saurabh W. Jha, Ken J. Shen, and Angus Williams for valuable input. D. Boubert thanks the UK Science and Technology Facilities Council for supporting his PhD. K. Hawkins thanks the Simons Society of Fellows and the Flatiron Institute Center for Computational Astrophysics. I. Ginsburg was supported in part by Harvard University funds and the Institute for Theory and Computation. J.S. acknowledges support from the Packard Foundation. This work has made use of data from the European Space Agency (ESA) mission Gaia(https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. Based on observations obtained at the Southern Astrophysical Research (SOAR) telescope, which is a joint project of the Ministério da Ciência, Tecnologia, Inovações e Comunicações (MCTIC) do Brasil, the U.S. National Optical Astronomy Observatory (NOAO), the University of North Carolina at Chapel Hill (UNC), and Michigan State University (MSU).

References

Appendix A The Open Fast Stars Catalog

The papers which originally proposed the late-type hypervelocity candidates discussed in the main text often give measurements of properties not generically included in Gaia, such as spectral types, radial velocities and other spectroscopic parameters. Combining these properties in a systematic, rigorous fashion with Gaia astrometry and photometry is crucial to determining the nature of these candidates. To that end, we have created the Open Fast Stars Catalog (OFSC) https://faststars.space utilizing the AstroCats framework (Guillochon et al. 2017). The objective of the catalogue is to contain a curated collection of every measurement of all high-velocity star candidates in the literature, with each measurement having a citable origin, and to utilize the available data to provide additional value to the community interested in these objects. At present, the OFSC has targeted the data available for stars that may potentially be hypervelocity stars, however we plan to expand it to include pulsars, runaway stars, and halo stars in the near future.

Like the other Open Astronomy Catalogs See https://astrocats.space, the OFSC adds value to the existent data by providing derived quantities to the community. The catalog automatically computes the amount of extinction to each object (see Section A.2), velocities in various frames (heliocentric, galactocentric), observability at a user-specified time from various observatory locations, probability of boundedness to the Milky Way, and correlations between observed and derived quantities. The catalog also provides an interface for each object with a near-complete collection of its data. At the moment, the catalog only includes fast star spectroscopy from the SDSS survey (Abolfathi et al. 2018) and the LAMOST survey (Luo et al. 2016), as little is available from public repositories; we plan to collect this data from the community in the near future.

In the subset of cases where we do not have either the proper motions or radial velocity then we assume that the missing component(s) exactly cancels the sampled solar reflex, which is equivalent to calculating the minimum Galactocentric rest-frame velocity. In this case the bound probability can be interpreted as an upper limit on the true bound probability.

A.2 Automatic querying of Gaia and other catalogues

A.3 The fast star graveyard

It is standard practise among the Open Astronomy Catalogs to split off objects which are no longer of interest, for instance transients falsely identified as supernovae are split off from the main Open Supernova Catalog. This practise is known as putting an object in the ‘graveyard’. In the OFSC, this can be interpreted as a statement that a fast star is highly unlikely to be unbound and thus should not be considered to be a hypervelocity candidate. The criteria for putting a star in the graveyard is that i) each of the six kinematic components have been measured, ii) the star has 5D astrometry from Gaia DR2, and iii) the star was bound in all of the $10^{4}$ samples. Note that a star being in the graveyard does not mean that it has been deleted and it will be possible for a star to be resurrected as new data is obtained, for instance when Gaia DR3 is released. The only practical result of a star being in the graveyard is that it is not shown in the main section of the OFSC. 159 previously-claimed hypervelocity candidate stars were in the OFSC graveyard as of 01/06/2018.