Astronomical Science

RAFT I: Discovery of New Planetary Candidates and Updated Orbits from Archival FEROS Spectra

Maritza G. Soto1  Figure 1. RV measurements James S. Jenkins1 3GHRVNQJ for HD 11977. 2 %$1.2 Matias I. Jones "'(1.-  ' 1/2

1 Departamento de Astronomía,   ­Universidad de Chile, Santiago, Chile R

2 Department of Electrical Engineering XL and Center of Astro-Engineering UC,  DKNBHS

Pontifica Universidad Católica de Chile, U Santiago, Chile

1@CH@K l

The first results of the Reanalysis l of Archival FEROS specTra (RAFT) pro- ject are presented. We have analysed FEROS data for five with pro- l           posed planetary companions in order )#l to test the reliability of the solutions with our new methodology. For HD 11977, HD 47536 and HD 110014 we confirm an operating spectral resolution of For this work, we have also used data the presence of one orbiting compan- R = 48 000 and a large wavelength range from three other instruments: HARPS ion. We reject the presence of a second (~ 350–920 nm). Several sys- (High Accuracy Planet companion around HD 47536, as well tems have been detected with FEROS, Searcher; Mayor et al., 2003), a spectro- as the planets detected around but some of these have been thrown into graph with which it is possible to reach HD 70573 and HD 122430. Finally, we doubt after rea­nalysis of the spectra. an accuracy on the order of 1 m s–1; propose the existence of a new second These differing results were obtained CORALIE (Queloz et al., 2000), for which planetary companion around the giant because of improvements in the calibra- it is possible to obtain an accuracy of HD 110014. tion technique, and by the discovery that ~ 6 m s–1, or better, for most of the ob­­ the FEROS pipeline barycentre correction servations; and CHIRON (Tokovinin et al., did not include a ­correction for the pre- 2013), with which it is possible to reach Exoplanetary science is a burgeoning cession of the coordinates (Müller et al., a precision in velocity of ~ 6 m s–1 (Jones field in astronomy, focusing on the detec- 2013). In this work the reanalysis of five et al., 2014). tion and study of planets outside the stars using FEROS is presented: HD 11977, Solar System. Indirect techniques are HD 47536, HD 70573, HD 110014 and In the following sections, the RAFT I mostly used to study , one of HD 122430, all of which have a reported results for the five stars are presented them being the radial velocity (RV) planet orbiting them. A more detailed and discussed. method, which can be described as fol- ­version of this work can be found in Soto lows: when a star has an orbiting com- et al. (2015). panion, it will move in a small orbit due to HD 11977 the gravitational pull produced by the All the data we analysed were obtained companion. It is possible to measure the from the ESO archive1. We reduced and This is a , with a of radial velocity of the star at different calibrated the spectra using the FEROS 2.31 MA), and a of –0.16 dex, epochs through the Doppler effect and pipeline, but disabled the barycentric with respect to Solar. A planet was dis- that way derive characteristics of the correction. To obtain the shift of the covered by Setiawan et al. (2005), with a companion, such as its minimum mass spectral lines, we cross-correlated the minimum mass of 6.5 Jupiter and , through the application spectra with a template (spectrum for (MJup) and an orbital period of 711 days. of simple Keplerian models. the star with a high signal-to-noise). We We reanalysed 48 spectra for this star then computed our own barycentric from FEROS, and also included 13 spec- The RV method uses spectrographs ­correction and applied it to the velocities. tra from HARPS and eight spectra taken to measure the spectrum of a star and Finally, we measured the internal drift, with CHIRON. We were able to detect retrieve the flux from the source as a which can be due to changes in pressure a period of 625 days in the data. Starting ­function of wavelength. After identifying and temperature within the instrument from that period, we minimised the orbital the spectral lines, we measure the enclosure, by measuring the velocity shift parameters of the system and found a ­velocity of the star as a function of the of the ­thorium–argon lines that are signal produced by a planet orbiting this time of observation. For this study, we observed simultaneously with the stellar star with a period of 621 days, a mini- used archival spectra from the Fibre-fed spectra, and we then removed this veloc- mum mass of 6.5 MJup and eccentricity of Extended Range Optical Spectrograph ity from the radial velocity measured for 0.3. The data, along with the fit, are (FEROS; Kaufer et al., 1999), which has each . shown in Figure 1. The precision of this

24 The Messenger 161 – September 2015   3GHRVNQJ 3GHRVNQJ  %$1.2 %$1.2 ' 1/2 ' 1/2 "'(1.-   ".1 +($

  R   R XL  L X DKNBHS

U l DKNBHS

U 

1@CH@K l 1@CH@K

l l

l

l l             )#l )#l

Figure 2. RV measurements for HD 47536. Figure 3. RV measurements for HD 110014. solution is 11.2 m s–1, significantly better improvement. We found that the proba- ond with a period of 130 days and a mini- than the one found previously, leading bility that the two-planet configuration is mum mass of 3.1 MJup. This fit is plotted us to believe these parameters for the similar to the one-planet one is ~ 70 %. in Figure 3. The scatter around this solu- system are more accurate. This shows that the two-planet solution is tion is 19.4 m s–1, significantly better than not a genuine configuration for the sys- the one-planet solution. We also found tem given the current data, (the data is that the probability that the two solutions HD 47536 being overfitted), and therefore it only are similar is essentially 0 %, meaning supports the existence of one planetary that this is the statistically preferred con-

This giant star has a mass of 0.98 MA companion. figuration. and a metallicity of –0.65 dex, making it the most metal-poor star in our sample. It was studied in two different publica- HD 110014 HD 122430 tions. In the first (Setiawan et al., 2003), a planet with a period of 619 days was This giant star has a mass of 2.09 MA This is a giant star with a mass of found, and later (Setiawan et al., 2008), and a metallicity of 0.14 dex, making it 1.68 MA and metallicity of –0.09 dex, after adding more radial velocities, a two- the star with the highest metallicity in our making it another metal-poor star. planet solution was reported: one signal sample. In de Medeiros et al. (2009) it Setiawan (2003) reported the discovery of with a period of 430 days, and the other was claimed that there is a planet orbiting a planet with a period of 344 days and with a period of 2500 days. the star with a mass of 9.5–11.1 MJup and minimum mass of 3.71–6.04 MJup. We a period of 835 days. We reanalysed 25 reanalysed the FEROS data for this star, We reanalysed 56 spectra from FEROS, FEROS spectra for this target, and also consisting of 42 radial velocities and also 18 from HARPS, 12 from CORALIE, and included 17 data points from HARPS. Our included six observations taken with six from CHIRON. We found a period periodogram search located a period of HARPS. We were not able to find any sig- of 434 days in the data, and starting from 833 days, and starting from this period nificant period in the data, so we tried to that we found a solution for a planet with we obtained a solution agreeing with a fit a solution for a planet with a starting a period of 434 days, a minimum mass planet of minimum mass 13.7 MJup and period equal to the one published before, of 4 MJup and an eccentricity of 0.3. The period of 936 days. This solution has a and then minimised the orbital parame- precision of this solution is 51.7 m s–1 scatter of 44.6 m s–1. ters. We finally obtained a solution for the and is shown in Figure 2. We tried to find orbit of a planet with a period of 455 days, another planet in this system, but were The residuals from this fit hinted at the minimum mass of 2.1 MJup, and eccen- not able to find any period that could pro- presence of a second signal with a period tricity of 0.6. The root mean square (RMS) vide a first solution. We did fit a planet of 133 days. When we tried to add a new scatter on the solution was 29.3 m s–1 with a period of 2500 days, obtaining a planet to the system with this starting and the fit is shown in Figure 4. Even solution with two planets and a precision period, we obtained a new configuration, though the scatter is smaller than we ob­­ of 48.9 m s–1. Although the two-planet consisting of two planets orbiting the tained for other fits, we are still not cer- solution has a better precision than the star: one with a period of 882 days and a tain about the existence of this compan- one-planet solution, it is not a great minimum mass of 10.7 MJup, and the sec- ion because of the lack of a significant

The Messenger 161 – September 2015 25 Astronomical Science Soto M. et al., RAFT I

period in the data. We conclude that  Figure 4. RV measure- there is not enough data to confirm the 3GHRVNQJ ments for HD 122430. %$1.2 presence of a companion around this star. ' 1/2 

HD 70573  R

L  This is a very young star, with an esti- X mated age of 78–125 Myrs, about 5 % of DKNBHS the Solar age, and a mass of 1.0 MA. Its KU rotational period was found to be 3.296 l days from photometry. A planet was 1@CH@ announced in Setiawan et al. (2007), with a mass of 6.1 MJup and an orbital period l of 852 days. We reanalysed 55 spectra taken with FEROS for this star and could find only one significant period in the l data. That period was 3.296 days,       remarkably similar to the star’s rotational )#l period. We fit a curve with this period and obtained a scatter of 36.4 m s–1. The Figure 5. Phased RV This work curve is shown in Figure 5. When we tried measurements for FEROS HD 70573. to fit the solution published before, we 100 obtained a scatter of 78.78 m s–1, much worse than our previous fit. We also found a period of 833 days, close to the ) 50 orbital period of the planet candidate, /s (m produced by the window function of the ty data. This means that the planet found loci 0 before could correspond to a signal pro- ve duced by the sampling of the data, and therefore would not be a signal produced Radial –50 by the presence of a planet orbiting this star. –100

Activity indices and photometric analysis –2.0 –1.5 –1.0 –0.5 0.010.51.0 .5 2.0 The shift produced by the spectral lines Phase in a star can be produced not only by the movement of said star, but also by velocities. Finally, we analysed the CCF detected companions. For example, if the phenomena in its , like dark full width at half maximum. More details signal from the proposed second com- spots on the stellar surface, or velocity about the use of these indices can be panion orbiting HD 110014 was produced fields produced by stellar convection found in numerous works, e.g., Dumusque by a dark spot on the surface of the star, zones. We computed three tests to verify et al. (2014), Jenkins et al. (2014), etc. For then it should cover 13.8 % of the sur- if our velocities were a product of these all the stars we found no significant cor- face. If that was the case, then a period phenomena. First, we calculated the relation between the radial velocities and of 130 days should be seen in the photo­ S-index, which measures the change in the activity indices for each epoch, which metry of HD 110014, but we do not find the flux from the Calcium II HK lines. led us to conclude that activity within the any. That led us to discard the possibility These are chromospheric lines that are stars is not the cause of the claimed that the 130- signal is produced by a affected by the magnetic activity within Doppler signals. rotating spot group on the stellar surface. the star. Then we performed a bisector analysis, which detects asymmetries We also studied the photometry of the The planets found in this work follow the in the line profiles caused by intrinsic stars for which we confirm a companion, trend that the higher the metallicity of a phenomena, by computing the bisector viz. HD 11977, HD 47536, and HD 110014. star, the higher the probability of detect- velocity span (BVS). This measures the We obtained the photometric data from ing an orbiting planet. This trend can be velocity difference between the bottom the All Sky Automated Survey (Pojmanski, seen in the case of HD 110014, which and top parts of the cross-correlation 1997). We found no periods in the data was the star with the highest metallicity function (CCF) used to measure the radial that could match the period of any in our sample and is the one with the

26 The Messenger 161 – September 2015 highest number of planets. One special space, leaving us with a sample of Setiawan, J. 2003, in Proc. Conf. on Towards case however is HD 47536. According around 80 stars that we plan to study Other Earths: DARWIN/TPF and the Search for Extrasolar Terrestrial Planets, 22–25 April 2003, to ­Reffert et al. (2015), there is almost 0 % in the future. ed. Fridlund,­ M., Henning, T. & Lacoste, H., probability of finding a planet orbiting a ­Heidelberg, Germany giant star with metallicity of –0.65 dex. The Setiawan, J. et al. 2003, A&A, 398, L19 existence of a planet around HD 47536 References Setiawan, J. et al. 2005, A&A, 437, L31 Setiawan, J. et al. 2007, ApJ, 660, L145 shows that it is possible to form planets de Medeiros, J. R. et al. 2009, A&A, 504, 617 Setiawan, J. et al. 2008, in Proc. ESO/Lisbon/Aveiro with masses on the order of the mass of Dumusque, X., Boisse, I. & Santos, N. C. 2014, Conference, 11–15 September 2006, ed. Santos, Jupiter around even very metal-poor stars. ApJ, 796, 132 N. C. et al., Garching, Germany, 325 Jenkins, J. S. & Tuomi, M. 2014, ApJ, 794, 110 Soto, M. G., Jenkins, J. S. & Jones, M. I. 2015, MNRAS, 451, 3131 We plan to continue with the RAFT pro- Jones, M. I. et al. 2014, A&A, 566, A113 Mayor, M. et al. 2003, The Messenger, 114, 20 Tokovinin, A. et al. 2013, PASP, 125, 1336 ject by reanalysing FEROS data for other Müller, A. et al. 2013, A&A, 556, A3 giant stars. Most of these stars belong Pojmanski, G. 1997, Acta Astron., 47, 467 to programmes aimed at detecting giant Queloz, D. et al. 2000, A&A, 354, 99 Links Reffert, S. et al. 2015, A&A, 574, A116 planets in this poorly studied parameter 1 ESO Archive: http://archive.eso.org ESO/Y. Beletsky

Night view over the La Silla Observatory from the north featuring the Galactic Plane. In the foreground is the Swedish ESO Submillimetre Telescope (SEST), which was decommissioned in 2003. The experi- ence gained on the SEST paved the way towards the Atacama Pathfinder Explorer (APEX) and the Atacama Large Millimeter/submillimeter Array (ALMA). In the background is the dome of the-3.6-metre telescope.

The Messenger 161 – September 2015 27