Exclusion of Stellar Companions to Exoplanet Host Stars

The discovery of exoplanets several decades ago (e.g., Latham et al. 1989; Wolszczan & Frail 1992; Latham 2012; Mayor & Queloz 1995) has led to a burgeoning and diverse field of study. A major effort of this work is directed at characterizing the individual exoplanets and their host stars. For example, determining the binarity of the host stars has become a crucial step in understanding exoplanetary systems, as the presence of a binarity companion can have a profound effect on detection methods and formation scenarios. This is particularly important as roughly half of all sun-like stars in the solar neighborhood are part of multiple-star systems (Raghavan et al. 2010; Horch et al. 2014). Indeed, the pursuit of Kepler candidates (Everett et al. 2015; Kraus et al. 2016) and Robo-AO observations of radial velocity (RV) exoplanet host stars has significantly contributed to our knowledge of this large rate of stellar multiplicity.

The presence of binary companions can considerably affect stellar measurements intended to discover and/or characterize exoplanets and cause severe blended contamination for transit observations (Cartier et al. 2015; Ciardi et al. 2015; Gilliland et al. 2015). One of the main consequences of this for exoplanets detected by the transit method is the underestimation of their planetary radii determined from the depth of the planetary transit (Ciardi et al. 2015; Furlan & Howell 2017). For systems discovered using the RV method, the presence of stellar companions can manifest as linear trends in the data, the precise origins of which can remain unresolved due to the insufficient time baseline and the $\sin i$ ambiguity of the companion mass interpretation (Crepp et al. 2012; Kane et al. 2014). The range of possible planetary formation scenarios is inhibited by the multiplicity of the stars and can impact such aspects as the orbital stability of the planets (Holman & Wiegert 1999). Furthermore, there has been a noted effect of stellar binarity on masses, orbital periods (Zucker & Mazeh 2002), and eccentricities (Eggenberger et al. 2004) of planets in such binary systems. Consequently, it has become critical for us to establish the multiplicity of exoplanet host stars so that we can be absolutely confident with the resulting interpretation of exoplanet signals and be successful in fully characterizing the overall system properties.

Wittrock et al. (2016) described a survey for exoplanet host star multiplicity and presented the detection of stellar companions to 2 of 71 surveyed exoplanet host stars, HD 2638 and HD 164509. Here, we present the results for the remaining 69 stars of the survey that place significant constraints on the presence of stellar companions to those stars. In Section 2, we discuss the method of detection, the range of targets that were selected for analysis, and the properties of null-detection systems. Section 3 briefly reviews the details of the data reduction and includes sensitivity plots of the observed systems. Section 4 presents the results from the data analysis, and Section 5 provides discussion of further work and concluding remarks.

A large survey project was established to search for stellar companions to a subset of the known RV exoplanet host stars. In total, 71 stars were observed in 2014 July using the Differential Speckle Survey Instrument, or DSSI (Horch et al. 2009); that instrument was stationed at the Gemini-north Observatory at the time of the observations. The stars were selected from the known RV exoplanet host star population where there was no known stellar companion. Two of the stars, HD 2638 and HD 164509, were found to have evidence for bound stellar companions contained in the data (Wittrock et al. 2016). Tables 1 and 2 list the 69 targets from the survey for which no stellar companion was detected. The first table includes spectral types, apparent magnitudes m_V, proper motions (denoted as μ), parallaxes, distances, and the number of exoplanets each star hosts, while the second table tallies stellar masses, radii, luminosity, effective temperatures, surface gravity, age, and metallicity. The data within both tables were taken from multiple literature and exoplanet databases (see reference section). The DSSI used two different filters, 692 and 880 nm, to acquire the speckle images of those targets. The 692 nm filter has FWHM of 40 nm, and the 880 nm filter has FWHM of 50 nm. All images were reduced using a data reduction pipeline, the details of which are provided in Section 3. Afterward, the images were examined by eye and also using the speckle reduced data plots for any companion source appearing next to the target.

Figure 1 showcases the Hertzsprung–Russell (H–R) diagram and distance histogram of the survey using the data from Tables 1 and 2. BD+48 738 and HD 13189 were excluded because their luminosity and/or age data are unavailable, and HD 240237 was excluded due to its extremely large distance of 5300 pc. As described by Wittrock et al. (2016), this is a magnitude-limited survey that targets the brightest of known exoplanet host stars, and so the sample consists mostly of relatively nearby dwarf stars with a peak in the distance distribution of ∼50 pc. The large distances of the giant stars results in a small angular separation sensitivity for detecting stellar companions.

Zoom In Zoom Out Reset image size

The details on the process undergone to obtain the final reconstructed images is provided in the previous paper (Wittrock et al. 2016) and with greater depth in Horch et al. (2012, 2015) but will be summarized here. The DSSI obtains the raw speckle data and stores it as FITS data cubes containing 1000 single short-exposure frames, where each frame is a 256 × 256-pixel image centered on the target. The plate scales for the observing run at Gemini-north were 0.01081 arcsec/pixel and 0.01120 arcsec/pixel for the 692 nm and 880 nm channels, respectively. The frames are bias-subtracted, autocorrelated, and then summed. The result is then Fourier transformed to retrieve the spatial frequency power spectrum of both the science target and a known unresolved point source standard. Afterward, the science target's power spectrum is divided by that of the point source to deconvolve the effects of the speckle transfer function and obtain a diffraction-limited estimate of the true power spectrum of the object. For the raw data frames, the image bispectrum of each frame has been created (more details on this process have been described in Lohmann et al. 1983). The relaxation algorithm of Meng et al. (1990) is then applied to calculate the phase of the object's Fourier transform. This result is then added with the square root of the deconvolved power spectrum to arrive at an estimate of the object's Fourier transform. Next, it is multiplied with a Gaussian low-pass filter of FWHM width equal to the telescope's diffraction limit. Lastly, we inverse-Fourier-transform the result to obtain the final reconstructed image.

With the reconstructed images in hand, we can use the method from Horch et al. (2011) to obtain a detection limit curve with respect to angular separation from the primary star. The average and standard deviation of the maxima inside annuli are computed to guide our estimation of the 5σ detection limit, which is the mean value plus five times the standard deviation. These values are in units of magnitude difference. The DSSI's diffraction limit on Gemini-north (0022 at 692 nm and 0027 at 880 nm) constrains the angular range of annuli from 01 to ∼12, and we arbitrarily chose an increment of 01 for the annuli. Afterward, we employed a cubic spline interpolation to achieve a smooth detection limit curve at all separations in between the two extreme limits. The sensitivity plots with these curves for some targets are listed in Figure 2. The construction of these sensitivity plots are described in more detail by Howell et al. (2011).

Zoom In Zoom Out Reset image size

Even though the remaining systems have not yielded the discovery of a stellar companion like those of HD 2638 and HD 164509 (Wittrock et al. 2016), such results nevertheless provide an important contribution to the stellar companion survey. Given that the field-of-view of the images is $2\buildrel{\prime\prime}\over{.} 8\times 2\buildrel{\prime\prime}\over{.} 8$, a null-detection implies that the system contains no stellar companions down to the sensitivity limit of the observation and within the projected size of the field-of-view. For example, one such system for which we do not detect a stellar companion and is relatively nearby (∼17 pc) is 14 Her (HD 145675). Wittenmyer et al. (2007) proposed a second exoplanet, 14 Her c, with a semimajor axis of 6.9 au and very low eccentricity of 0.02 ± 0.06. Our non-detection of a stellar companion around 14 Her out to ∼25 au (see Table 3) indicates that the observed signal for the outer body is highly likely to be caused by a planetary body.

Table 3.  Limiting Magnitudes

Name	Exclusion Radius (au)		5σ Δm Limit (692 nm)		5σ Δm Limit (880 nm)		m − M	Max Mass (M⊙)
	Inner	Outer	01	02	01	02		692 nm	880 nm
BD+14 4559	2.42	67.70	3.92	4.32	3.93	4.62	3.42	0.26	0.26
BD+48 738	17.54	491.23	3.09	5.01	3.72	4.74	7.73	1.35	1.14
GJ 581	0.31	8.70	4.36	5.14	4.29	5.23	−1.03	0.09	0.09
GJ 649	0.52	14.48	3.40	4.23	3.39	4.27	0.07	0.17	0.17
GJ 849	0.45	12.73	4.01	4.55	3.95	4.55	−0.21	0.13	0.13
HD 1461	1.16	32.54	3.21	4.15	3.54	4.89	1.83	0.45	0.41
HD 1502	7.96	222.93	4.44	5.20	4.29	4.97	6.01	0.63	0.65
HD 2638	2.50	69.90	4.45	4.96	3.83	4.63	3.49	0.24	0.28
HD 3651	0.55	15.48	4.53	5.25	4.31	5.15	0.22	0.25	0.27
HD 4313	6.85	191.78	4.67	5.39	4.28	5.21	5.68	0.62	0.69
HD 5319	5.72	160.18	4.09	5.30	4.15	5.23	5.29	0.62	0.61
HD 5891	12.56	351.76	4.41	5.37	4.43	5.31	7.00	0.89	0.89
HD 6718	2.74	76.80	4.32	4.80	4.20	4.95	3.70	0.33	0.34
HD 7449	1.95	54.50	4.28	4.91	4.21	4.80	2.95	0.35	0.35
HD 8574	2.23	62.39	4.53	5.18	4.45	5.06	3.24	0.39	0.40
HD 9446	2.62	73.30	4.75	5.24	4.49	5.08	3.59	0.29	0.31
HD 10697	1.63	45.60	4.50	5.20	4.10	4.80	2.56	0.41	0.46
HD 12661	1.75	48.93	4.51	5.22	4.16	5.15	2.72	0.32	0.35
HD 13189	28.09	786.52	4.03	5.33	3.73	4.87	8.75	0.00	0.00
HD 13931	2.21	61.92	3.55	5.05	3.84	4.97	3.23	0.44	0.41
HD 16175	2.89	81.02	3.44	5.03	4.08	5.30	3.81	0.57	0.48
HD 16400	4.63	129.51	2.28	4.22	2.99	3.66	4.83	1.77	1.47
HD 16760	2.27	63.64	3.93	4.99	3.81	4.77	3.29	0.30	0.31
HD 17092	5.43	152.17	3.11	4.64	4.27	5.07	5.18	1.30	0.96
HD 136118	2.33	65.21	4.49	5.37	4.07	5.07	3.34	0.42	0.47
HD 136418	4.91	137.52	4.03	4.99	3.59	4.52	4.96	0.60	0.67
HD 137510	2.06	57.76	4.27	4.92	3.84	4.83	3.08	0.49	0.55
HD 139357	5.90	165.29	2.83	4.73	3.74	4.65	5.36	1.62	1.28
HD 142245	5.48	153.34	4.30	5.01	4.73	5.30	5.20	0.67	0.60
HD 143107	3.39	95.04	4.33	5.32	4.05	5.25	4.16	1.34	1.44
HD 143107	3.39	95.04	4.44	5.19	4.26	5.20	4.16	1.30	1.37
HD 143761	0.86	24.13	2.65	3.66	4.17	4.96	1.18	0.58	0.39
HD 143761	0.86	24.13	4.09	4.81	3.72	4.68	1.18	0.40	0.44
HD 143761	0.86	24.13	4.42	4.95	4.33	5.19	1.18	0.36	0.37
HD 145457	6.27	175.44	4.44	4.96	4.62	5.15	5.49	0.90	0.86
HD 145675	0.88	24.60	4.30	5.13	4.18	5.34	1.22	0.28	0.29
HD 149143	3.10	86.85	4.39	4.96	4.12	5.09	3.96	0.39	0.42
HD 152581	9.28	259.74	4.52	5.09	4.53	5.05	6.34	0.67	0.67
HD 154345	0.93	26.02	3.70	4.96	3.60	4.72	1.35	0.33	0.34
HD 155358	2.21	61.76	3.37	4.75	3.34	4.55	3.22	0.51	0.51
HD 156279	1.83	51.24	2.93	4.53	4.22	4.86	2.82	0.42	0.30
HD 156668	1.22	34.26	3.90	5.18	3.53	4.71	1.94	0.25	0.27
HD 158038	5.18	145.08	4.48	5.24	4.22	5.07	5.08	0.62	0.67
HD 163607	3.44	96.35	3.30	4.51	4.40	4.95	4.19	0.55	0.41
HD 164509	2.62	73.41	3.90	4.15	4.00	4.52	3.60	0.39	0.38
HD 164922	1.11	30.97	3.69	4.88	3.64	4.76	1.72	0.34	0.35
HD 167042	2.51	70.32	3.22	3.84	4.29	5.15	3.50	0.84	0.64
HD 170693	4.83	135.14	2.58	4.05	2.74	3.84	4.92	2.10	2.02
HD 171028	5.49	153.85	3.84	4.09	4.00	4.45	5.20	0.54	0.52
HD 173416	6.97	195.26	3.33	3.51	3.75	4.01	5.72	1.45	1.30
HD 177830	2.95	82.64	3.13	3.23	4.05	4.29	3.86	0.71	0.56
HD 180314	6.57	183.97	3.23	3.48	4.13	4.46	5.59	1.23	0.97
HD 187123	2.41	67.57	3.89	3.98	3.93	4.34	3.42	0.40	0.39
HD 190228	3.08	86.26	3.82	4.03	4.07	4.56	3.95	0.56	0.52
HD 192263	0.97	27.04	1.85	3.55	3.34	4.21	1.43	0.44	0.29
HD 197037	1.62	45.26	4.07	4.25	4.05	4.63	2.55	0.39	0.39
HD 199665	3.77	105.42	3.82	4.30	3.89	4.48	4.38	1.01	0.99
HD 200964	3.61	101.08	1.97	3.18	3.70	4.60	4.29	1.23	0.78
HD 206610	9.69	271.32	4.15	4.60	4.52	5.00	6.44	0.77	0.69
HD 208527	20.16	564.52	4.02	4.17	3.84	4.10	8.03	2.18	2.29
HD 210277	1.08	30.19	4.14	4.27	4.42	5.05	1.67	0.33	0.30
HD 210702	2.75	76.92	4.12	4.54	4.48	4.74	3.70	0.70	0.64
HD 217014	0.78	21.85	4.07	4.60	3.49	4.24	0.97	0.37	0.43
HD 217107	0.99	27.80	4.23	4.88	4.35	5.27	1.49	0.34	0.33
HD 217786	2.74	76.80	4.51	5.20	4.44	5.16	3.70	0.37	0.37
HD 218566	1.43	39.98	4.41	5.19	4.20	5.18	2.28	0.22	0.23
HD 219828	3.62	101.23	4.41	5.19	4.31	5.12	4.30	0.42	0.43
HD 220074	16.23	454.55	3.21	3.89	4.31	4.67	7.56	2.58	1.93
HD 220773	2.54	71.25	4.25	4.94	4.22	4.87	3.53	0.45	0.46
HD 221345	3.96	110.85	4.43	5.00	3.93	4.87	4.49	0.98	1.12
HD 222155	2.45	68.69	3.91	4.15	4.24	4.61	3.45	0.48	0.44
HD 231701	5.92	165.88	4.10	4.18	4.10	4.31	5.37	0.46	0.46
HD 240210	7.14	200.00	3.72	4.13	3.72	4.25	5.77	1.46	1.46
HD 240237	263.16	7368.42	3.30	4.65	4.52	5.18	13.61	0.23	0.17

Download table as:  ASCIITypeset images: 1 2

Table 3 includes the inner and outer exclusion radii, limiting magnitudes at $0\buildrel{\prime\prime}\over{.} 1$ and $0\buildrel{\prime\prime}\over{.} 2$, the distance moduli, and the maximum stellar mass of a hypothetical companion for each target, including HD 2638 and HD 164509 from Wittrock et al. (2016). The exclusion radii are the range of physical separations from the host star that are observable within the Gemini DSSI's field-of-view. The exclusion radii and the distance moduli are calculated using the stellar distances provided in Table 1. The minimum angular separation is constrained by Gemini's diffraction limits of about $0\buildrel{\prime\prime}\over{.} 022$ at 692 nm and $0\buildrel{\prime\prime}\over{.} 027$ at 880 nm and is $0\buildrel{\prime\prime}\over{.} 05$. As mentioned before, the maximum angular separation is $1\buildrel{\prime\prime}\over{.} 2$, which provides a constraint on the outer exclusion radius. Thus, the exclusion radii provide a region where stellar companions with certain spectral types may be excluded. Therefore, the last two columns tell us the maximum mass a stellar companion can have before becoming detectable via DSSI within the exclusion region. These masses were calculated using a simple mass–luminosity relationship of ${({L}_{\star }/{L}_{\odot })=({M}_{\star }/{M}_{\odot })}^{3.5}$ (Kuiper 1938) and the given limiting magnitudes at both 692 and 880 nm.

We constructed three plots, shown in Figures 3 and 4, using the information from Table 3. Figure 3 shows some correlation between the inner exclusion radius and effective temperature of the targets, with tight clusters at or below 5 au. The large inner exclusion radius for some of the cooler stars represents the giant stars in the sample, as verified by the distance indicators. The plots displayed in Figure 4 show that the limiting magnitudes of our observations are largely consistent between the two passbands used of 692 and 880 nm. As for Figure 1, BD+48 738, HD 13189, and HD 240237 were excluded from the plots (see Section 2).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Detection of stellar companions to exoplanet host stars is now relatively common. Their discovery, or lack of it, are of beneficial contributions to the structure of stellar systems, as the presence of a stellar companion significantly influences the orbital dynamics and models for formation processes. For planets discovered using the RV technique, the search for stellar companions plays a major role in the correct interpretation of residual RV trends present in the data. Furthermore, determining if any of the known exoplanet host stars are single, binary, or multiple systems is absolutely essential in avoiding situations where a gas giant would be mistaken for a terrestrial or even an Earth-like planet.

Our DSSI survey monitored 71 stars for which 2 were detected to have stellar companions (Wittrock et al. 2016), and the remaining 69 show no evidence of stellar companions within the $2\buildrel{\prime\prime}\over{.} 8\times 2\buildrel{\prime\prime}\over{.} 8$ field-of-view and above the instrument's sensitivity limit. This increases the probability that remaining detected objects in the RV data, if any, are planetary bodies if not extremely low-mass stars or brown dwarfs. An example of this is the 14 Her system, for which the partial phase coverage of the RV signal detected by Wittenmyer et al. (2007) is better explained by a planetary body rather than stellar, as our exclusion range out to ∼25 au completely encompasses the postulated semimajor axis of the proposed 14 Her c. The exclusion radii listed in Table 3 provide a physical range of separations from each star within which any future RV detection of objects gravitationally bound to these stars may now be more closely associated with a planetary object.

The authors would like to thank the referee for providing feedback that improved the quality of the paper. This work is based on observations obtained at the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Research Council (Canada), CONICYT (Chile), the Australian Research Council (Australia), Ministério da Ciência, Tecnologia e Inovação (Brazil) and Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina). This research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program. The results reported herein benefited from collaborations and/or information exchange within NASA's Nexus for Exoplanet System Science (NExSS) research coordination network sponsored by NASA's Science Mission Directorate.