Abstract

We explore the possibility of planet formation in the carbon-rich protoplanetary discs of carbon-enhanced metal-poor (CEMP) stars, possible relics of the early Universe. The chemically anomalous abundance patterns ([C/Fe] ≥ 0.7) in this subset of low-mass stars suggest pollution by primordial core-collapsing supernovae ejecta that are particularly rich in carbon dust grains. By comparing the dust-settling time-scale in the protoplanetary discs of CEMP stars to the expected disc lifetime (assuming dissipation via photoevaporation), we determine the maximum distance rmax from the host CEMP star at which carbon-rich planetesimal formation is possible, as a function of the host star's [C/H] abundance. We then use our linear relation between rmax and [C/H], along with the theoretical mass–radius relation derived for a solid, pure carbon planet, to characterize potential planetary transits across host CEMP stars. Given that the related transits are detectable with current and upcoming space-based transit surveys, we suggest initiating an observational programme to search for carbon planets around CEMP stars in hopes of shedding light on the question of how early planetary systems may have formed after the big bang.

1 INTRODUCTION

The questions of when, where, and how the first planetary systems formed in cosmic history remain crucial to our understanding of structure formation and the emergence of life in the early Universe (Loeb 2014). In the cold dark matter model of hierarchical structure formation, the first stars are predicted to have formed in dark matter haloes that collapsed at redshifts z ≲ 50, about 100 million years after the big bang (Tegmark et al. 1997; Barkana & Loeb 2001; Yoshida et al. 2003; Bromm & Larson 2004; Loeb & Furlanetto 2013). These short-lived, metal-free, massive first-generation stars ultimately exploded as supernovae (SNe) and enriched the interstellar medium (ISM) with the heavy elements fused in their cores. The enrichment of gas with metals that had otherwise been absent in the early Universe enabled the formation of the first low-mass stars, and perhaps, marked the point at which star systems could begin to form planets (Bromm & Loeb 2003; Frebel, Johnson & Bromm 2007; Clark, Glover & Klessen 2008). In the core accretion model of planet formation (e.g. Papaloizou & Terquem 2006; Janson et al. 2011), elements heavier than hydrogen and helium are necessary not only to form the dust grains that are the building blocks of planetary cores, but to extend the lifetime of the protostellar disc long enough to allow the dust grains to grow via merging and accretion to form planetesimals (Kornet et al. 2005; Johansen, Youdin & Mac Low 2009; Yasui et al. 2009; Ercolano & Clarke 2010).

In the past four decades, a broad search has been launched for low-mass Population II stars in the form of extremely metal-poor sources within the halo of the Galaxy. The HK survey (Beers, Preston & Shectman 1985, 1992), the Hamburg/ESO Survey (Wisotzki et al. 1996; Christlieb et al. 2008), the Sloan Digital Sky Survey (SDSS; York et al. 2000), and the Sloan Extension for Galactic Understanding and Expolration survey (Yanny et al. 2009) have all significantly enhanced the sample of metal-poor stars with [Fe/H] < −2.0. Although these iron-poor stars are often referred to in the literature as ‘metal-poor’ stars, it is critical to note that [Fe/H] does not necessarily reflect a stellar atmosphere's total metal content. The equivalence between ‘metal-poor’ and ‘Fe-poor’ appears to fall away for stars with [Fe/H] < −3.0 since many of these stars exhibit large overabundances of elements such as C, N, and O; the total mass fractions, Z, of the elements heavier than He are therefore not much lower than the solar value in these iron-poor stars.

Carbon-enhanced metal-poor (CEMP) stars comprise one such chemically anomalous class of stars, with carbon-to-iron ratios [C/Fe] ≥ 0.7 (as defined in Aoki et al. 2007; Carollo et al. 2012; Norris et al. 2013). The fraction of sources that fall into this category increases from ∼15–20 per cent for stars with [Fe/H] < −2.0, to 30 per cent for [Fe/H] < −3.0, to ∼75 per cent for [Fe/H] < −4.0 (Beers & Christlieb 2005; Norris et al. 2013; Frebel & Norris 2015). Furthermore, the degree of carbon enhancement in CEMP stars has been shown to notably increase as a function of decreasing metallicity, rising from [C/Fe] ∼ 1.0 at [Fe/H] = −1.5 to [C/Fe] ∼ 1.7 at [Fe/H] = −2.7 (Carollo et al. 2012). Given the significant frequency and level of carbon-excess in this subset of metal-poor Population II stars, the formation of carbon planets around CEMP stars in the early Universe presents itself as an intriguing possibility.

From a theoretical standpoint, the potential existence of carbon exoplanets, consisting of carbides and graphite instead of Earth-like silicates, has been suggested by Kuchner & Seager (2005). Using the various elemental abundances measured in planet-hosting stars, subsequent works have sought to predict the corresponding variety of terrestrial exoplanet compositions expected to exist (Bond, O'Brien & Lauretta 2010; Carter-Bond et al. 2012a; Carter-Bond, O'Brien & Raymond 2012b). Assuming that the stellar abundances are similar to those of the original circumstellar disc, related simulations yield planets with a whole range of compositions, including some that are almost exclusively C and SiC; these occur in discs with C/O > 0.8, favourable conditions for carbon condensation (Larimer 1975). Observationally, there have also been indications of planets with carbon-rich atmospheres, e.g. WASP-12b (Madhusudhan et al. 2011), and carbon-rich interiors, e.g. 55 Cancri e (Madhusudhan, Lee & Mousis 2012).

In this paper, we explore the possibility of carbon planet formation around the iron-deficient, but carbon-rich subset of low-mass stars, mainly, CEMP stars. In Section 2, we discuss the origins of the unique elemental abundance patterns among these C-rich objects and their potential implications for the carbon dust content of the gas from which CEMP stars and their protostellar discs form. Comparing the expected disc lifetime to the dust-settling time-scale in these protostellar discs, we then determine the maximum distance from a host CEMP star at which the formation of a carbon-rich planet is possible (Section 3). In Section 4, we calculate the theoretical mass–radius relation for such a pure carbon planet and present the corresponding depth and duration of its transit across the face of its host CEMP star in Section 5. We conclude with a discussion of our findings in Section 6. Standard definitions of elemental abundances and ratios are adopted in this paper. For element X, the logarithmic absolute abundance is defined as the number of atoms of element X per 1012 hydrogen atoms, log ϵ(X) = log10(NX/NY) + 12.0. For elements X and Y, the logarithmic abundance ratio relative to the solar ratio is defined as [X/Y] = log10(NX/NY) − log10(NX/NY). The solar abundance set is that of Asplund et al. (2009), with a solar metallicity Z = 0.0134.

2 STAR-FORMING ENVIRONMENT OF CEMP STARS

A great deal of effort has been directed in the literature towards understanding theoretically, the origin of the most metal-poor stars, and in particular, the large fraction that is C-rich. These efforts have been further perturbed by the fact that CEMP stars do not form a homogenous group, but can rather be further subdivided into two main populations (Beers & Christlieb 2005): carbon-rich stars that show an excess of heavy neutron-capture elements (CEMP-s, CEMP-r, and CEMP-r/s), and carbon-rich stars with a normal pattern of the heavy elements (CEMP-no). In the following sections, we focus on stars with [Fe/H] ≤ −3.0, which have been shown to fall almost exclusively in the CEMP-no subset (Aoki 2010).

A number of theoretical scenarios have been proposed to explain the observed elemental abundances of these stars, though there is no universally accepted hypothesis. The most extensively studied mechanism to explain the origin of CEMP-no stars is the mixing and fallback model, where a ‘faint’ Population III SN explodes, but due to a relatively low explosion energy, only ejects its outer layers, rich in lighter elements (up to magnesium); its innermost layers, rich in iron and heavier elements, fall back on to the remnant and are not recycled in the ISM (Umeda & Nomoto 2003, 2005). This potential link between primeval SNe and CEMP-no stars is supported by recent studies which demonstrate that the observed ratio of carbon-enriched to carbon-normal stars with [Fe/H] < −3.0 is accurately reproduced if SNe were the main source of metal-enrichment in the early Universe (Cooke & Madau 2014; de Bennassuti et al. 2014). Furthermore, the observed abundance patterns of CEMP-no stars have been found to be generally well matched by the nucleosynthetic yields of primordial faint SNe (Iwamoto et al. 2005; Umeda & Nomoto 2005; Tominaga, Umeda & Nomoto 2007; Joggerst, Woosley & Heger 2009; Yong et al. 2013; Ishigaki et al. 2014; Keller et al. 2014; Marassi et al. 2014, 2015; Tominaga, Iwamoto & Nomoto 2014; Bonifacio et al. 2015). These findings suggest that most of the CEMP-no stars were probably born out of gas enriched by massive, first-generation stars that ended their lives as Type II SNe with low levels of mixing and a high degree of fallback.

Under such circumstances, the gas clouds which collapse and fragment to form these CEMP-no stars and their protostellar discs may contain significant amounts of carbon dust grains. Observationally, dust formation in SNe ejecta has been inferred from isotopic anomalies in meteorites, where graphite, SiC, and Si3N4 dust grains have been identified as SNe condensates (Zinner 1998). Furthermore, in situ dust formation has been unambiguously detected in the expanding ejecta of SNe such as SN 1987A (Lucy et al. 1989; Indebetouw et al. 2014) and SN 1999em (Elmhamdi et al. 2003). The existence of cold dust has also been verified in the SN remnant of Cassiopeia A by Submillimetre Bolometer Array's recent submillimetre observations, and a few solar masses worth of dust is estimated to have condensed in the ejecta (Dunne et al. 2003).

Theoretical calculations of dust formation in primordial core-collapsing SNe have demonstrated the condensation of a variety of grain species, starting with carbon, in the ejecta, where the mass fraction tied up in dust grains grows with increasing progenitor mass (Kozasa, Hasegawa & Nomoto 1989; Todini & Ferrara 2001; Nozawa et al. 2003). Marassi et al. (2014, 2015) consider, in particular, dust formation in weak Population III SNe ejecta, the type believed to have polluted the birth clouds of CEMP-no stars. Tailoring the SN explosion models to reproduce the observed elemental abundances of CEMP-no stars, they find that: (i) for all the progenitor models investigated, amorphous carbon (AC) is the only grain species that forms in significant amounts; this is a consequence of extensive fallback, which results in a distinct, carbon-dominated ejecta composition with negligible amounts of other metals, such as Mg, Si, and Al, that can enable the condensation of alternative grain types; (ii) the mass of carbon locked into AC grains increases when the ejecta composition is characterized by an initial mass of C greater than the O mass; this is particularly true in zero metallicity SN progenitors, which undergo less mixing than their solar metallicity counterparts (Joggerst et al. 2009); in their stratified ejecta, C-grains are found only to form in layers where C/O > 1; in layers where C/O < 1, all the carbon is promptly locked in CO molecules; (iii) depending on the model, the mass fraction of dust (formed in SNe ejecta) that survives the passage of an SN reverse shock ranges between 1 and 85 per cent; this fraction is referred to as the carbon condensation efficiency; (iv) further grain growth in the collapsing birth clouds of CEMP-no stars, due to the accretion of carbon atoms in the gas phase on to the remaining grains, occurs only if C/O > 1 and is otherwise hindered by the formation of CO molecules.

Besides the accumulation of carbon-rich grains imported from the SNe ejecta, Fischer–Trope-type reactions (FTTs) may also contribute to solid carbon enrichment in the protostellar discs of CEMP-no stars by enabling the conversion of nebular CO and H2 to other forms of carbon (Llorca & Casanova 1998; Kress & Tielens 2001). Furthermore, in carbon-rich gas, the equilibrium condensation sequence changes significantly from the sequence followed in solar composition gas where metal oxides condense first. In nebular gas with C/O ≳ 1, carbon-rich compounds such as graphite, carbides, nitrides, and sulfides are the highest temperature condensates [T ≈ 1200–1600 K] (Larimer 1975). Thus, if planet formation is to proceed in this C-rich gas, the protoplanetary discs of these CEMP-no stars may spawn many carbon planets.

3 ORBITAL RADII OF POTENTIAL CARBON PLANETS

Given the significant abundance of carbon grains, both imported from SNe ejecta and produced by equilibrium and non-equilibrium mechanisms operating in the C-rich protoplanetary discs, the emerging question is: would these dust grains have enough time to potentially coagulate and form planets around their host CEMP-no stars?

In the core accretion model, terrestrial planet formation is a multistep process, starting with the aggregation and settling of dust grains in the protoplanetary disc (Lissauer 1993; Beckwith, Henning & Nakagawa 2000; Papaloizou & Terquem 2006; Nagasawa et al. 2007; Rieke 2008; Armitage 2010; Janson et al. 2011). In this early stage, high densities in the disc allow particles to grow from submicron-size to metre-size through a variety of collisional processes including Brownian motion, settling, turbulence, and radial migration. The continual growth of such aggregates by coagulation and sticking eventually leads to the formation of kilometre-sized planetesimals, which then begin to interact gravitationally and grow by pairwise collisions, and later by runaway growth (Weidenschilling 1988; Lissauer 1993; Weidenschilling & Cuzzi 1993). In order for terrestrial planets to ultimately form, these processes must all occur within the lifetime of the disc itself, a limit which is set by the relevant time-scale of the physical phenomena that drive disc dissipation.

A recent study by Yasui et al. (2009) of clusters in the extreme outer Galaxy provides observational evidence that low-metallicity discs have shorter lifetimes (<1 Myr) compared to solar metallicity discs (∼5–6 Myr). This finding is consistent with models in which photoevaporation by energetic (ultraviolet or X-ray) radiation of the central star is the dominant disc dispersal mechanism. While the opacity source for extreme-ultraviolet (EUV) photons is predominantly hydrogen and is thus metallicity independent, X-ray photons are primarily absorbed by heavier elements, mainly carbon and oxygen, in the inner gas and dust shells. Therefore, in low-metallicity environments where these heavy elements are not abundant and the opacity is reduced, high-density gas at larger columns can be ionized and will experience a photoevaporative flow if heated to high enough temperatures (Gorti & Hollenbach 2009; Ercolano & Clarke 2010).

Assuming that photoevaporation is the dominant mechanism through which circumstellar discs lose mass and eventually dissipate, we adopt the metallicity dependent disc lifetime, derived in Ercolano & Clarke (2010) using X-ray+EUV models (Ercolano, Clarke & Drake 2009),
\begin{equation} t_{{\rm disc}}\propto Z^{0.77(4-2p)/(5-2p)}, \end{equation}
(1)
where Z is the total metallicity of the disc and p is the power-law index of the disc surface density profile (Σ ∝ rp). A mean power-law exponent of p ∼ 0.9 is derived by modelling the spatially resolved emission morphology of young stars at (sub)millimetre wavelengths (Andrews et al. 2009, 2010) and the time-scale is normalized such that the mean lifetime for discs of solar metallicity is 2 Myr (Ercolano & Clarke 2010). We adopt the carbon abundance relative to solar [C/H] as a proxy for the overall metallicity Z since the opacity, which largely determines the photoevaporation rate, and thus the disc lifetime, is dominated by carbon dust grains in the CEMP-no stars we consider in this paper.
The time-scale for planet formation is believed to be effectively set by the time it takes dust grains to settle into the disc mid-plane. The subsequent process of runaway planetesimal formation, possibly occurring via a series of pairwise collisions, must be quick, since otherwise, the majority of the solid disc material would radially drift towards the host star and evaporate in the hot inner regions of the circumstellar disc (Armitage 2010). We adopt the one-particle model of Dullemond & Dominik (2005) to follow the mass growth of dust grains via collisions as they fall through and sweep up the small grains suspended in the disc. Balancing the gravitational force felt by a small dust particle at height z above the mid-plane of a disc with the aerodynamic drag (in the Epstein regime) gives a dust settling velocity of
\begin{equation} v_{{\rm sett}}=\frac{{\rm d}z}{{\rm d}t}=\frac{3\Omega _{\rm K}^2zm}{4\rho c_{\rm s}\sigma _{\rm d}} \end{equation}
(2)
where σd = πa2 is the cross-section of the dust grain with radius a and |$c_{\rm s}=\sqrt{k_BT(r)/\mu m_{\rm H}}$| is the isothermal sound speed with mH being the mass of a hydrogen atom and μ = 1.36 being the mean molecular weight of the gas (including the contribution of helium). |$\Omega _{\rm K}=\sqrt{GM_{\ast }/r^3}$| is the Keplerian velocity of the disc at a distance r from the central star of mass M*, which we take to be M* = 0.8 M as representative of the low masses associated with CEMP-no stars (Christlieb et al. 2002; Frebel & Norris 2015). The disc is assumed to be in hydrostatic equilibrium with a density given by
\begin{equation} \rho (z,r)=\frac{\Sigma (r)}{h\sqrt{2\pi }}\exp {\left(-\frac{z^2}{2h^2}\right)}, \end{equation}
(3)
where the disc scaleheight is h = csk. For the disc surface density Σ(r) and temperature T(r) profiles, we adopt the radial power-law distributions fitted to (sub)millimetre observations of circumstellar discs around young stellar objects (Andrews & Williams 2005; Andrews et al. 2009, 2010),
\begin{equation} T(r)=200 \textrm{ K} \left(\frac{r}{1 \textrm{ AU}}\right)^{-0.6} \end{equation}
(4)
\begin{equation} \Sigma (r)=10^3 \,\mathrm{ g\,cm^{-2}}\left(\frac{r}{1 \textrm{ AU}}\right)^{-0.9} . \end{equation}
(5)
Although these relations were observationally inferred from discs with solar-like abundances, we choose to rely on them for our purposes given the lack of corresponding measurements for discs around stars with different abundance patterns.
The rate of grain growth, dm/dt, is determined by the rate at which grains, subject to small-scale Brownian motion, collide and stick together as they drift towards the disc mid-plane through a sea of smaller solid particles. If coagulation results from every collision, then the mass growth rate of a particle is effectively the amount of solid material in the volume swept out the particle's geometric cross-section,
\begin{equation} \frac{{\rm d}m}{{\rm d}t}=f_{{\rm dg}}\rho \sigma _{\rm d}\left(v_{{\rm rel}}+\frac{{\rm d}z}{{\rm d}t}\right), \end{equation}
(6)
where dz/dt is the dust-settling velocity given by equation (2) and
\begin{equation} v_{{\rm rel}}=\sqrt{\frac{8k_BT(m_1+m_2)}{\pi m_1m_2}}\approx \sqrt{\frac{8k_BT}{\pi m}} \end{equation}
(7)
is the relative velocity in the Brownian motion regime between grains with masses m1 = m2 = m. To calculate the dust-to-gas mass ratio in the disc fdg, we follow the approach in Ji, Frebel & Bromm (2014) and relate two expressions for the mass fraction of C: (i) the fraction of carbon in the dust, fdgMC, dust/Mdust, where Mdust is the total dust mass and MC, dust is the carbon dust mass ; and (ii) the fraction of carbon in the gas, μCnCnH, where μC is the molecular weight of carbon (∼12mp) and nC and nH are the carbon and hydrogen number densities, respectively.
We then assume that a fraction fcond (referred to from now on as the carbon condensation efficiency) of all the carbon present in the gas cloud is locked up in dust, such that
\begin{equation} f_{{\rm cond}}\frac{\mu _{\rm C} n_{\rm C}}{\mu n_{\rm H}}=f_{{\rm dg}}\frac{M_{{\rm C,dust}}}{M_{{\rm dust}}} \,\,. \end{equation}
(8)
Since faint Population III SNe are believed to have polluted the birth clouds of CEMP-no stars, and the only grain species that forms in non-negligible amounts in these ejecta is AC (Marassi et al. 2014, 2015), we set Mdust = MC, dust. Rewriting equation (8) in terms of abundances relative to the Sun, we obtain
\begin{equation} f_{{\rm dg}}=f_{{\rm cond}}\frac{\mu _{\rm C}}{\mu }10^{\textrm{[C/H]}+\log {\epsilon ({\rm C})_{\odot }}-12} \end{equation}
(9)
where log ϵ(C) = 8.43±0.05 (Asplund et al. 2009) is the solar carbon abundance.

For a specified metallicity [C/H] and radial distance r from the central star, we can then estimate the time it takes for dust grains to settle in the disc by integrating equations (2) and (6) from an initial height of z(t = 0) = 4h with an initial dust grain mass of |$m(t=0)=4\pi a_{{\rm init}}^3\rho _{\rm d}/3$|⁠. The specific weight of dust is set to ρd = 2.28 g cm−3, reflecting the material density of carbon grains expected to dominate the circumstellar discs of CEMP-no stars. The initial grain size ainit is varied between 0.01 and 1 μm to reflect the range of characteristic radii of carbon grains found when modelling CEMP-no star abundance patterns (Marassi et al. 2014). Comparing the resulting dust-settling time-scale to the disc lifetime given by equation (1) for the specified metallicity, we can then determine whether there is enough time for carbon dust grains to settle in the mid-plane of the disc and there undergo runaway planetesimal formation before the disc is dissipated by photoevaporation. For the purposes of this simple model, we neglected possible turbulence in the disc which may counteract the effects of vertical settling, propelling particles to higher altitudes and thus preventing them from fully settling into the disc mid-plane (Armitage 2010). We have also not accounted for the effects of radial drift, which may result in the evaporation of solid material in the hot inner regions of the circumstellar disc.

As the dust-settling time-scale is dependent on the disc surface density Σ(r) and temperature T(r), we find that for a given metallicity, [C/H], there is a maximum distance rmax from the central star out to which planetesimal formation is possible. At larger distances from the host star, the dust settling time-scale exceeds the disc lifetime and so carbon planets with semimajor axes r > rmax are not expected to form. A plot of the maximum semimajor axis expected for planet formation around a CEMP-no star as a function of the carbon abundance relative to the Sun [C/H] is shown in Fig. 1 for carbon condensation efficiencies ranging between fcond = 0.1 and 1. As discovered in Johnson & Li (2012), where the critical iron abundance for terrestrial planet formation is considered as a function of the distance from the host star, we find a linear relation between [C/H] and rmax,
\begin{equation} \textrm{[C/H]}=\log {\left(\frac{r_{{\rm max}}}{1 \textrm{ au}}\right)}-\alpha , \end{equation}
(10)
where α = 1.3, 1.7, and 1.9 for fcond = 0.1, 0.5, and 1, respectively, assuming an initial grain size of ainit = 0.1 μm. These values for α change by less than 1 per cent for smaller initial grain sizes, ainit = 0.01 μm, and by no more than 5 per cent for larger initial grain sizes ainit = 1 μm; given this weak dependence on ainit, we only show our results for a single initial grain size of ainit = 0.1 μm. The distance from the host star at which the temperature of the disc approaches the sublimation temperature of carbon dust, Tsub, C ∼ 2000 K (Kobayashi et al. 2011), is depicted as well (dash–dotted grey curve). At distances closer to the central star than r ≃ 0.02 au, temperatures well exceed the sublimation temperature of carbon grains; grain growth and subsequent carbon planetesimal formation are therefore quenched in this inner region.
Figure 1.

The maximum distance rmax from the host star out to which planetesimal formation is possible as a function of the star's metallicity, expressed as the carbon abundance relative to that of the Sun, [C/H]. The dotted, dashed, and solid black curves correspond to the results obtained assuming carbon condensation efficiencies of 10 per cent, 50 per cent, and 100 per cent, respectively, and an initial grain size of ainit = 0.1 μm. The grey dash–dotted curve corresponds to the distance at which the disc temperature approaches the sublimation temperature of carbon dust grains, Tsub, C ∼ 2000 K; the formation of carbon planetesimals will therefore be suppressed at distances that fall below this line, r ≲ 0.02 au. The coloured vertical lines represent various observed CEMP stars with measured carbon abundances, [C/H].

Fig. 1 shows lines representing various observed CEMP stars with measured carbon abundances, mainly, HE 0107-5240 (Christlieb et al. 2002, 2004), SDSS J0212+0137 (Bonifacio et al. 2015), SDSS J1742+2531 (Bonifacio et al. 2015), G 77-61 (Dahn et al. 1977; Plez & Cohen 2005; Beers et al. 2007), and HE 2356-0410 (Norris, Ryan & Beers 1997; Roederer et al. 2014). These stars all have iron abundances (relative to solar) [Fe/H] < −3.0, carbon abundances (relative to solar) [C/Fe] > 2.0, and carbon-to-oxygen ratios C/O > 1. This latter criteria maximizes the abundance of solid carbon available for planet formation in the circumstellar discs by optimizing carbon grain growth both in stratified SNe ejecta and later, in the collapsing molecular birth clouds of these stars. It also advances the possibility of carbon planet formation by ensuring that planet formation proceeds by a carbon-rich condensation sequence in the protoplanetary disc. SDSS J0212+0137 and HE 2356-0410 have both been classified as CEMP-no stars, with measured barium abundances [Ba/Fe] < 0 (as defined in Beers & Christlieb 2005); the other three stars are Ba-indeterminate, with only high upper limits on [Ba/Fe], but are believed to belong to the CEMP-no subclass given their light-element abundance patterns. The carbon abundance, [C/H], dominates the total metal content of the stellar atmosphere in these five CEMP objects, contributing more than 60 per cent of the total metallicity in these stars. A summary of the relevant properties of the CEMP stars considered in this analysis can be found in Table 1. We find that carbon planets may be orbiting iron-deficient stars with carbon abundances [C/H] ∼ −0.6, such as HE 2356-0410, as far out as ∼20 au from their host star in the case where fcond = 1. Planets forming around stars with less carbon enhancement, i.e. HE 0107-5240 with [C/H] ∼ −1.6, are expected to have more compact orbits, with semimajor axes r < 2 au. If the carbon condensation efficiency is only 10 per cent, the expected orbits grow even more compact, with maximum semimajor axes of ∼ 5 and 0.5 au, respectively.

Table 1.

Basic dataa for CEMP stars considered in this paper.

Starlog gb[Fe/H][C/Fe]C/OcSourced
HE 0107-5240 2.2 −5.44 3.82 14.1 1, 2 
SDSS J0212+0137 4.0 −3.57 2.26 2.6 
SDSS J1742+2531 4.0 −4.77 3.60 2.2 
G 77-61 5.1 −4.03 3.35 12.0 4, 5 
HE 2356-0410e 2.65 −3.19 2.61 >14.1 
Starlog gb[Fe/H][C/Fe]C/OcSourced
HE 0107-5240 2.2 −5.44 3.82 14.1 1, 2 
SDSS J0212+0137 4.0 −3.57 2.26 2.6 
SDSS J1742+2531 4.0 −4.77 3.60 2.2 
G 77-61 5.1 −4.03 3.35 12.0 4, 5 
HE 2356-0410e 2.65 −3.19 2.61 >14.1 

aAbundances based on one-dimensional local thermodynamic equilibrium model-atmosphere analyses.

bLogarithm of the gravitational acceleration at the surface of stars expressed in cm s−2.

cC/O = |$N_{\mathrm{C}}/N_{\mathrm{O}} = 10^{\mathrm{[C/O]}+\log {\epsilon (\mathrm{C})_{\odot }}-\log {\epsilon (\mathrm{O})_{\odot }}}$|⁠.

dReferences: (1) Christlieb et al. (2004); (2) Collet, Asplund & Trampedach (2006); (3)Bonifacio et al. (2015); (4) Plez & Cohen (2005); (5) Beers et al. (2007); (6)Roederer et al. (2014).

eCS 22957-027

Table 1.

Basic dataa for CEMP stars considered in this paper.

Starlog gb[Fe/H][C/Fe]C/OcSourced
HE 0107-5240 2.2 −5.44 3.82 14.1 1, 2 
SDSS J0212+0137 4.0 −3.57 2.26 2.6 
SDSS J1742+2531 4.0 −4.77 3.60 2.2 
G 77-61 5.1 −4.03 3.35 12.0 4, 5 
HE 2356-0410e 2.65 −3.19 2.61 >14.1 
Starlog gb[Fe/H][C/Fe]C/OcSourced
HE 0107-5240 2.2 −5.44 3.82 14.1 1, 2 
SDSS J0212+0137 4.0 −3.57 2.26 2.6 
SDSS J1742+2531 4.0 −4.77 3.60 2.2 
G 77-61 5.1 −4.03 3.35 12.0 4, 5 
HE 2356-0410e 2.65 −3.19 2.61 >14.1 

aAbundances based on one-dimensional local thermodynamic equilibrium model-atmosphere analyses.

bLogarithm of the gravitational acceleration at the surface of stars expressed in cm s−2.

cC/O = |$N_{\mathrm{C}}/N_{\mathrm{O}} = 10^{\mathrm{[C/O]}+\log {\epsilon (\mathrm{C})_{\odot }}-\log {\epsilon (\mathrm{O})_{\odot }}}$|⁠.

dReferences: (1) Christlieb et al. (2004); (2) Collet, Asplund & Trampedach (2006); (3)Bonifacio et al. (2015); (4) Plez & Cohen (2005); (5) Beers et al. (2007); (6)Roederer et al. (2014).

eCS 22957-027

4 MASS–RADIUS RELATIONSHIP FOR CARBON PLANETS

Next, we present the relationship between the mass and radius of carbon planets that we have shown may theoretically form around CEMP-no stars. These mass–radius relations have already been derived in the literature for a wide range of rocky and icy exoplanet compositions (Zapolsky & Salpeter 1969; Léger et al. 2004; Valencia, O'Connell & Sasselov 2006; Fortney, Marley & Barnes 2007; Seager et al. 2007). Here, we follow the approach of Zapolsky & Salpeter (1969) and solve the three canonical equations of internal structure for solid planets,

  • mass conservation
    \begin{equation} \frac{{\rm d}m(r)}{{\rm d}r}=4\pi r^2 \rho (r) \,, \end{equation}
    (11)
  • hydrostatic equilibrium
    \begin{equation} \frac{{\rm d}P(r)}{{\rm d}r}=-\frac{Gm(r)\rho (r)}{r^2} \,,\,\,\mathrm{and} \end{equation}
    (12)
  • the equation of state (EOS)
    \begin{equation} P(r) = f\left(\rho (r),T(r)\right) \,, \end{equation}
    (13)

where m(r) is the mass contained within radius r, P(r) is the pressure, ρ(r) is the density of the spherical planet, and f is the unique EOS of the material of interest, in this case, carbon.

Carbon grains in circumstellar discs most likely experience many shock events during planetesimal formation which may result in the modification of their structure. The coagulation of dust into clumps, the fragmentation of the disc into clusters of dust clumps, the merging of these clusters into ∼ 1 km planetesimals, the collision of planetesimals during the accretion of meteorite parent bodies, and the subsequent collision of the parent bodies after their formation all induce strong shock waves that are expected to chemically and physically alter the materials (Mimura & Sugisaki 2003). Subject to these high temperatures and pressures, the AC grains polluting the protoplanetary discs around CEMP stars are expected to undergo graphitization and may even crystallize into diamond (Tielens et al. 1987; Onodera, Higashi & Irie 1988; Papoular et al. 1996; Takai et al. 2003). In our calculations, the EOS at low pressures, P ≤ 14 GPa, is set to the third-order finite strain Birch–Murnagham EOS (BME; Birch 1947; Poirier 2000) for graphite,
\begin{equation} P=\frac{3}{2}K_0\left(\eta ^{7/3}-\eta ^{5/3}\right)\left[1+\frac{3}{4}\left(K_0^{\prime }-4\right)\left(\eta ^{2/3}-1\right)\right], \end{equation}
(14)
where η = ρ/ρ0 is the compression ratio with respect to the ambient density, ρ0, K0 is the bulk modulus of the material, and |$K_0^{\prime }$| is the pressure derivative. Empirical fits to experimental data yields a BME EOS of graphite (ρ0 = 2.25 g cm−3) with parameters K0 = 33.8 GPa and |$K_0^{\prime }$| = 8.9 (Hanfland, Beister & Syassen 1989). At 14 GPa, we incorporate the phase transition from graphite to diamond (Naka et al. 1976; Hanfland et al. 1989) and adopt the Vinet EOS (Vinet et al. 1987, 1989),
\begin{equation} P=3K_0\eta ^{2/3}\left(1-\eta ^{-1/3}\right)\exp {\left[\frac{3}{2}\left(K_0^{\prime }-1\right)\left(1-\eta ^{-1/3}\right)\right]} \end{equation}
(15)
with K0 = 444.5 GPa and |$K_0^{\prime }$| = 4.18 empirically fit for diamond, ρ0 = 3.51 g cm−3 (Dewaele et al. 2008). (As pointed out in Seager et al. (2007), the BME EOS is not fit to be extrapolated to high pressures since it is derived by expanding the elastic potential energy as a function of pressure keeping only the lowest order terms.) Finally, at pressures P ≳ 1300 GPa where electron degeneracy becomes increasingly important, we use the Thomas–Fermi–Dirac theoretical EOS (Salpeter & Zapolsky 1967; equations 40– 49), which intersects the diamond EOS at P ∼ 1300 GPa. Given that the full temperature-dependent carbon EOSs are either undetermined or dubious at best, all three EOSs adopted in this work are room-temperature EOSs for the sake of practical simplification.

Using a fourth-order Runge–Kutta scheme, we solve the system of equations simultaneously, numerically integrating equations (11) and (12) beginning at the planet's centre with the inner boundary conditions M(r = 0) = 0 and P(r = 0) = Pcentral, where Pcentral is the central pressure. The outer boundary condition P(r = Rp) = 0 then defines the planetary radius Rp and total planetary mass Mp = m(r = Rp). Integrating these equations for a range of Pcentral, with the appropriate EOS, P = P(ρ), to close the system of equations, yields the mass–radius relationship for a given composition. We show this mass–radius relation for a purely solid carbon planet in Fig. 2. We find that for masses Mp ≲ 800 M, gravitational forces are small compared with electrostatic Coulomb forces in hydrostatic equilibrium and so the planet's radius increases with increasing mass, |$R_{\rm p} \propto M_{\rm p}^{1/3}$|⁠. However, at larger masses, the electrons are pressure-ionized and the resulting degeneracy pressure becomes significant, causing the planet radius to become constant and even decrease for increasing mass, |$R_{\rm p} \propto M_{\rm p}^{-1/3}$| (Hubbard 1984). Planets which fall within the mass range 500 ≲ Mp ≲ 1300 M, where the competing effects of Coulomb forces and electron degeneracy pressure cancel each other out, are expected to be approximately the same size, with Rp ≃ 4.3 R, the maximum radius of a solid carbon planet. (In the case of gas giants, the planet radius can increase due to accretion of hydrogen and helium.)

Figure 2.

Mass–radius relation for solid homogenous, pure carbon planet.

Although the mass–radius relation illustrated in Fig. 2 may alone not be enough to confidently distinguish a carbon planet from a water or silicate planet, the unique spectral features in the atmospheres of these carbon planets may provide the needed fingerprints. At high temperatures (T ≳ 1000 K), the absorption spectra of massive (M ∼ 10–60 M) carbon planets are expected to be dominated by CO, in contrast with the H2O-dominated spectra of hot massive planets with solar-composition atmospheres (Kuchner & Seager 2005). The atmospheres of low-mass (M ≲ 10 M) carbon planets are also expected to be differentiable from their solar-composition counterparts due to their abundance of CO and CH4, and lack of oxygen-rich gases like CO2, O2, and O3 (Kuchner & Seager 2005). Furthermore, carbon planets of all masses at low temperatures are expected to accommodate hydrocarbon synthesis in their atmospheres; stable long-chain hydrocarbons are therefore another signature feature that can help observers distinguish the atmospheres of cold carbon planets and more confidently determine the bulk composition of a detected planet (Kuchner & Seager 2005).

5 TRANSIT PROPERTIES

The detection of theoretically proposed carbon planets around CEMP stars will provide us with significant clues regarding how early planet formation may have started in the Universe. While direct detection of these extrasolar planets remains difficult given the low luminosity of most planets, techniques such as the transit method are often employed to indirectly spot exoplanets and determine physical parameters of the planetary system. When a planet ‘transits’ in front of its host star, it partially occludes the star and causes its observed brightness to drop by a minute amount. If the host star is observed during one of these transits, the resulting dip in its measured light curve can yield information regarding the relevant sizes of the star and the planet, the orbital semimajor axis, and the orbital inclination, among other characterizing properties.

A planetary transit across a star is characterized by three main parameters: the fractional change in the stellar brightness, the orbital period, and the duration of the transit (Borucki et al. 1996). The fractional change in brightness is referred to as the transit depth, ΔF (with a total observed flux F), and is simply defined as the ratio of the planet's area to the host star's area (Seager & Mallén-Ornelas 2003),
\begin{equation} \Delta F = \frac{F_{\mathrm{no transit}}-F_{\mathrm{transit}}}{F_{\mathrm{no transit}}}=\left(\frac{R_{\rm p}}{R_{\ast }}\right)^2 \,. \end{equation}
(16)
Given the stellar radius R*, measurements of the relative flux change ΔF yield estimates of the size of the planet Rp, and the corresponding planetary mass Mp if the mass–radius relation for the planet is known. Using the RpMp relation derived in Section 4, we illustrate in Fig. 3 how the transit depth varies as a function of planetary mass in the case where a pure carbon planet transits across the face of its host CEMP star. Curves are shown for each of the five CEMP stars considered in this study, where we assume a stellar mass of M* = 0.8 M (representative of the low masses associated with old, iron-poor stellar objects) and derive the stellar radii using the stellar surface gravities g listed in Table 1, |$R_{\ast } = \sqrt{GM_{\ast }/g}$|⁠.
Figure 3.

Transit depth as a function of the planetary mass (for a solid carbon planet transit) for different host stellar radii with an assumed stellar mass of M* = 0.8 M. These particular stellar radii were chosen to correspond to the stellar radii of the CEMP stars considered in this paper, mainly HE 0107-5240 (red), G 77-61(cyan), HE 2356-0410 (blue), SDSS J0212+0137, and SDSS J1742+2531 (magenta), where the last two CEMP objects have the same measured surface gravity log g = 4.0.

As can be seen in Fig. 3, the relative change in flux caused by an Earth-mass carbon planet transiting across its host CEMP star ranges from ∼0.0001 per cent for a host stellar radius of R* ∼ 10 R to ∼0.01 per cent for a solar-sized stellar object. These shallow transit depths are thus expected to evade detection by ground-based transit surveys, which are generally limited in sensitivity to fractional flux changes on the order of 0.1 per cent (Beichman, Greene & Krist 2009). To push the limits of detection down to smaller, low-mass terrestrial planets requires space-based transit surveys that continuously monitor a large number of potential host stars over several years and measure their respective transit light curves. There are a number of ongoing, planned, and proposed space missions committed to this cause, including CoRot (COnvection ROtation and planetary Transits), Kepler, PLATO (PLAnetary Transits and Oscillations of stars), TESS (Transiting Exoplanet Survey Satellite), and ASTrO (All Sky Transit Observer), which are expected to achieve precisions as low as 20–30 ppm [parts per million] (Beichman et al. 2009; Winn & Fabrycky 2015). With the ability to measure transit depths as shallow as ΔF ∼ 0.001 per cent, these space transit surveys offer a promising avenue towards detecting the planetary systems that may have formed around CEMP stars.

The orbital period of a planet P, which can be determined if consecutive transits are observed, is given by Kepler's third law in the case of a circular orbit,
\begin{equation} P^2=\frac{4\pi ^2 a^3}{G(M_{\ast }+M_{\rm p})}\simeq \frac{4\pi ^2 a^3}{GM_{\ast }}, \end{equation}
(17)
where a is the orbital semimajor axis and the planetary mass is assumed to be negligible relative to the stellar mass, MpM*, in the second equality. Given the relation we derived in equation (10) between the metallicity [C/H] and the maximum semimajor axis allowed for a planet orbiting a CEMP star, the maximum orbital period of the planet can be expressed as a function of the metallicity of the host CEMP star (M* = 0.8 M),
\begin{equation} P_{\mathrm{max}}\simeq 365.25 \frac{10^{\frac{3}{2}(\mathrm{[C/H]}+\alpha )}}{\sqrt{M_{\ast }/\mathrm{M}_{\odot }}} \mathrm{ d}, \end{equation}
(18)
where α ≃ 1.3, 1.7, and 1.9 for carbon condensation efficiencies of 10 per cent, 50 per cent, and 100 per cent. As can be seen in Fig. 4, CEMP stars with higher carbon abundances [C/H], i.e. G 77-61 and HE 2356-0410, and larger efficiencies for carbon dust condensation fcond, can host planets with wider orbits and slower rotations, resulting in transits as infrequently as once every couple hundred years. Conversely, carbon planets orbiting relatively less carbon-rich CEMP stars, like SDSS J0212+0137, are expected to have higher rates of transit reoccurrence, completing rotations around their parent stars every ∼ 1–10 yr. These shorter period planets therefore have a much higher probability of producing an observable transit.
Figure 4.

Maximum orbital period of a carbon planet transiting across its host CEMP star as a function of the star's metallicity, expressed as the carbon abundance relative to that of the Sun, [C/H]. The dotted, dashed, and solid black curves denote the results obtained assuming carbon condensation efficiencies of 10 per cent, 50 per cent, and 100 per cent in the parent CEMP star with mass M* = 0.8 M. The coloured vertical lines represent the five CEMP stars considered in this paper with measured carbon abundances [C/H].

The maximum duration of the transit, T, can also be expressed as a function of the metallicity [C/H] of the parent CEMP star. For transits across the centre of a star, the total duration is given by,
\begin{equation} T\simeq 2R_{\ast }\sqrt{\frac{a}{GM_{\ast }}} \end{equation}
(19)
with the assumption that MpM* and RpR*. Once again, using the relation from equation (10) to express the maximum orbital distance from the host star in terms of the star's carbon abundance, we find that the maximum transit duration of a carbon planet across its parent CEMP star (M* = 0.8 M) is
\begin{equation} T_{{\rm max}}\simeq 13\,\frac{R_{\ast }}{\mathrm{R}_{\odot }}\sqrt{\frac{10^{\mathrm{[C/H]}+\alpha }}{M_{\ast }/\mathrm{M}_{\odot }}} \mathrm{ h}\,\,. \end{equation}
(20)
In Fig. 5, these maximum durations are shown as a function of [C/H] for the various stellar radii associated with the CEMP stars we consider in this paper. Transits across CEMP stars with larger radii and higher carbon abundances are expected to take much longer. While the total transit duration across SDSS J0212+0137 and SDSS J1742+2531 with R* ∼ 1.5 R and metallicities of [C/H] ∼ −1.3 – −1.2, is at most ∼ 1–2 d, transits across HE 0107-5240 (R* ∼ 12 R, [C/H]∼ −1.6) can take up to 2 weeks (fcond = 1). In general, the geometric probability of a planet passing between the observer and the planet's parent star increases with stellar radius and decreases with orbital radius, ptR*/a (Kane 2007). Therefore, focusing on CEMP stars, such as HE 0107-5240 and HE 2356-0410, with large stellar radii increases the observer's chance of spotting transits and detecting a planetary system.
Figure 5.

Maximum total transit duration of a carbon planet across its host CEMP star as a function of the star's metallicity, expressed as the carbon abundance relative to that of the Sun, [C/H]. The dotted, dashed, and solid black curves denote the results obtained assuming carbon condensation efficiencies of 10 per cent, 50 per cent, and 100 per cent. in the parent CEMP star. The coloured vertical lines represent the five CEMP stars considered in this paper with measured carbon abundances [C/H] and stellar radii derived using |$R_{\ast }=\sqrt{GM_{\ast }/g}$| with mass M* = 0.8 M

6 DISCUSSION

We explored in this paper, the possibility of carbon planet formation around the iron-deficient, carbon-rich subset of low-mass stars known as CEMP stars. The observed abundance patterns of CEMP-no stars suggest that these stellar objects were probably born out of gas enriched by massive first-generation stars that ended their lives as Type II SNe with low levels of mixing and a high degree of fallback. The formation of dust grains in the ejecta of these primordial core-collapsing SNe progenitors has been observationally confirmed and theoretically studied. In particular, AC is the only grain species found to condense and form in non-negligible amounts in SN explosion models that are tailored to reproduce the abundance patterns measured in CEMP-no stars. Under such circumstances, the gas clouds which collapse and fragment to form CEMP-no stars and their protoplanetary discs may contain significant amounts of carbon dust grains imported from SNe ejecta. The enrichment of solid carbon in the protoplanetary discs of CEMP stars may then be further enhanced by FTTs and carbon-rich condensation sequences, where the latter occurs specifically in nebular gas with C/O ≳ 1.

For a given metallicity [C/H] of the host CEMP star, the maximum distance out to which planetesimal formation is possible can then be determined by comparing the dust-settling time-scale in the protostellar disc to the expected disc lifetime. Assuming that disc dissipation is driven by a metallicity dependent photoevaporation rate, we find a linear relation between [C/H] and the maximum semimajor axis of a carbon planet orbiting its host CEMP star. Very carbon-rich CEMP stars, such as G 77-61 and HE 2356-0410 with [C/H] ≃ −0.7 – −0.6, can host carbon planets with semimajor axes as large ∼ 20 au for 100 per cent carbon condensation efficiencies; this maximum orbital distance reduces to ∼ 5 au when the condensation efficiency drops by an order of magnitude. In the case of the observed CEMP-no stars HE 0107-5240, SDSS J0212+0137, and SDSS J1742+2531, where the carbon abundances are in the range [C/H] ≃ −1.6–−1.2, we expect more compact orbits, with maximum orbital distances rmax ≃ 2, 4, and 6 au, respectively, for fcond = 1 and rmax ≃ 0.5–1 au for fcond = 0.1.

We then use the linear relation found between [C/H] and rmax (Section 3), along with the theoretical mass–radius relation derived for a solid, pure carbon planet (Section 4), to compute the three observable characteristics of planetary transits: the orbital period, the transit depth, and the transit duration. We find that the relative change in flux, ΔF, caused by an Earth-mass carbon planet transiting across its host CEMP star ranges from ∼ 0.0001 per cent for a stellar radius of R* ∼ 10 R to ∼ 0.01 per cent for a solar-sized stellar host. While the shallow transit depths of Earth-mass carbon planets around HE 0107-5240 and HE 2356-0410 may evade detection, current and future space-based transit surveys promise to achieve the precision levels (ΔF ∼ 0.001 per cent) necessary to detect planetary systems around CEMP stars such as SDSS J0212+0137, SDSS J1742+2531, and G 77-61.

Short orbital periods and long transit durations are also key ingredients in boosting the probability of transit detection by observers. G 77-61 is not an optimal candidate in these respects since given its large carbon abundance ([C/Fe] ∼ 3.4), carbon planets may form out to very large distances and take up to a century to complete an orbit around the star for fcond = 1 (Pmax ∼ 10 yr for 10 per cent carbon condensation efficiency). The small stellar radius, R* ∼ 0.5 R, also reduces chances of spotting the transit since the resulting transit duration is only ∼ 30 h at most. Carbon planets around larger CEMP stars with an equally carbon-rich protoplanetary disc, such as HE 2356-0410 (R* ∼ 7 R), have a better chance of being spotted, with transit durations lasting up to ∼ 3 weeks. The CEMP-stars SDSS J0212+0137 and SDSS J1742+2531 are expected to host carbon planets with much shorter orbits, Pmax ∼ 16 yr for 100 per cent condensation efficiency (Pmax ∼ 1 yr for fcond = 0.1), and transit durations that last as long as ∼60 h. If the ability to measure transit depths improves to a precision of 1 ppm, then potential carbon planets around HE 0107-5240 are the most likely to be spotted (among the group of CEMP-no stars considered in this paper), transiting across the host star at least once every ∼5 months (10 per cent condensation efficiency) with a transit duration of 6 d.

While our calculations place upper bounds on the distance from the host star out to which carbon planets can form, we note that orbital migration may alter a planet's location in the circumstellar disc. As implied by the existence of ‘hot Jupiters’, it is possible for a protoplanet that forms at radius r to migrate inward either through gravitational interactions with other protoplanets, resonant interactions with planetesimals with more compact orbits, or tidal interactions with gas in the surrounding disc (Papaloizou & Terquem 2006). Since Fig. 1 only plots rmax, the maximum distance out to which a carbon planet with [C/H] can form, our results remain consistent in the case of an inward migration. However, unless planets migrate inward from their place of birth in the disc, we do not expect to find carbon exoplanets orbiting closer than r ≃ 0.02 au from the host stars since at such close proximities, temperatures are high enough to sublimate carbon dust grains.

Protoplanets can also be gravitationally scattered into wider orbits through interactions with planetesimals in the disc (Hahn & Malhotra 1999; Veras, Crepp & Ford 2009). Such an outward migration of carbon planets may result in observations that are inconsistent with the curves in Fig. 1. A planet that formed at radius rrmax still has room to migrate outwards without violating the ‘maximum distance’ depicted in Fig. 1; however, the outward migration of a carbon planet that originally formed at, or near, rmax would result in a breach of the upper bounds placed on the transiting properties of carbon planets (Section 5). In particular, a carbon planet that migrates to a semimajor axis r > rmax will have an orbital period and a transit duration time that exceeds the limits prescribed in equations (18) and (20), respectively.

Detection of the carbon planets that we suggest may have formed around CEMP stars will provide us with significant clues regarding how planet formation may have started in the early Universe. The formation of planetary systems not only signifies an increasing degree of complexity in the young Universe, but it also carries implications for the development of life at this early junction (Loeb 2014). The lowest metallicity planetary system detected to date is around BD+20 24 57, a K2-giant with [Fe/H] = −1.0 (Niedzielski et al. 2009), a metallicity already well below the critical value once believed to be necessary for planet formation (Gonzalez, Brownlee & Ward 2001; Pinotti et al. 2005). More recent formulations of the minimum metallicity required for planet formation are consistent with this observation, estimating that the first Earth-like planets likely formed around stars with metallicities [Fe/H] ≲ −1.0 (Johnson & Li 2012). The CEMP stars considered in this paper are extremely iron-deficient, with [Fe/H] ≲ −3.2, and yet, given the enhanced carbon abundances which dominate the total metal content in these stars ([C/H] ≳ −1.6), the formation of solid carbon exoplanets in the protoplanetary discs of CEMP stars remains a real possibility. An observational programme aimed at searching for carbon planets around these low-mass Population II stars could therefore potentially shed light on the question of how early planets, and subsequently, life could have formed after the big bang.

We are thankful to Sean Andrews and Karin Öberg for helpful discussions and feedback. This work was supported in part by NSF grant AST-1312034. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE1144152. Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

REFERENCES

Andrews
S. M.
Williams
J. P.
2005
ApJ
631
1134

Andrews
S. M.
Wilner
D. J.
Hughes
A. M.
Qi
C.
Dullemond
C. P.
2009
ApJ
700
1502

Andrews
S. M.
Wilner
D. J.
Hughes
A. M.
Qi
C.
Dullemond
C. P.
2010
ApJ
723
1241

Aoki
W.
2010
Cunha
K.
Spite
M.
Barbuy
B.
Proc. IAU Symp 265, Chemical Abundances in the Universe: Connecting First Stars to Planets
Cambridge Univ. Press
Cambridge
111

Aoki
W.
Beers
T. C.
Christlieb
N.
Norris
J. E.
Ryan
S. G.
Tsangarides
S.
2007
ApJ
655
492

Armitage
P. J.
2010
Astrophysics of Planet Formation
Cambridge Univ. Press
Cambridge

Asplund
M.
Grevesse
N.
Sauval
A. J.
Scott
P.
2009
ARA&A
47
481

Barkana
R.
Loeb
A.
2001
Phys. Rep.
349
125

Beckwith
S. V. W.
Henning
T.
Nakagawa
Y.
2000
Mannings
V.
Boss
A. P.
Russell
S. S.
Protostars and Planets IV
Univ. Arizona Press
Tuscan, AZ
533

Beers
T. C.
Christlieb
N.
2005
ARA&A
43
531

Beers
T. C.
Preston
G. W.
Shectman
S. A.
1985
AJ
90
2089

Beers
T. C.
Preston
G. W.
Shectman
S. A.
1992
AJ
103
1987

Beers
T. C.
Sivarani
T.
Marsteller
B.
Lee
Y.
Rossi
S.
Plez
B.
2007
AJ
133
1193

Beichman
C. A.
Greene
T.
Krist
J.
2009
Pont
F.
Sasselov
D.
Holman
M.
Proc. IAU Symp. 253, Transiting Planets
Cambridge Univ. Press
Cambridge
319

Birch
F.
1947
Phys. Rev.
71
809

Bond
J. C.
O'Brien
D. P.
Lauretta
D. S.
2010
ApJ
715
1050

Bonifacio
P.
et al. 
2015
A&A
579
A28

Borucki
W. J.
Dunham
E. W.
Koch
D. G.
Cochran
W. D.
Rose
J. D.
Cullers
D. K.
Granados
A.
Jenkins
J. M.
1996
Ap&SS
241
111

Bromm
V.
Larson
R. B.
2004
ARA&A
42
79

Bromm
V.
Loeb
A.
2003
Nature
425
812

Carollo
D.
et al. 
2012
ApJ
744
195

Carter-Bond
J. C.
O'Brien
D. P.
Delgado Mena
E.
Israelian
G.
Santos
N. C.
González Hernández
J. I.
2012a
ApJ
747
L2

Carter-Bond
J. C.
O'Brien
D. P.
Raymond
S. N.
2012b
ApJ
760
44

Christlieb
N.
et al. 
2002
Nature
419
904

Christlieb
N.
Gustafsson
B.
Korn
A. J.
Barklem
P. S.
Beers
T. C.
Bessell
M. S.
Karlsson
T.
Mizuno-Wiedner
M.
2004
ApJ
603
708

Christlieb
N.
Schörck
T.
Frebel
A.
Beers
T. C.
Wisotzki
L.
Reimers
D.
2008
A&A
484
721

Clark
P. C.
Glover
S. C. O.
Klessen
R. S.
2008
ApJ
672
757

Collet
R.
Asplund
M.
Trampedach
R.
2006
ApJ
644
L121

Cooke
R. J.
Madau
P.
2014
ApJ
791
116

Dahn
C. C.
Liebert
J.
Kron
R. G.
Spinrad
H.
Hintzen
P. M.
1977
ApJ
216
757

de Bennassuti
M.
Schneider
R.
Valiante
R.
Salvadori
S.
2014
MNRAS
445
3039

Dewaele
A.
Datchi
F.
Loubeyre
P.
Mezouar
M.
2008
Phys. Rev. B
77
094106

Dullemond
C. P.
Dominik
C.
2005
A&A
434
971

Dunne
L.
Eales
S.
Ivison
R.
Morgan
H.
Edmunds
M.
2003
Nature
424
285

Elmhamdi
A.
et al. 
2003
MNRAS
338
939

Ercolano
B.
Clarke
C. J.
2010
MNRAS
402
2735

Ercolano
B.
Clarke
C. J.
Drake
J. J.
2009
ApJ
699
1639

Fortney
J. J.
Marley
M. S.
Barnes
J. W.
2007
ApJ
659
1661

Frebel
A.
Norris
J. E.
2015
ARA&A
53
631

Frebel
A.
Johnson
J. L.
Bromm
V.
2007
MNRAS
380
L40

Gonzalez
G.
Brownlee
D.
Ward
P.
2001
Icarus
152
185

Gorti
U.
Hollenbach
D.
2009
ApJ
690
1539

Hahn
J. M.
Malhotra
R.
1999
AJ
117
3041

Hanfland
M.
Beister
H.
Syassen
K.
1989
Phys. Rev. B
39
12598

Hubbard
W. B.
1984
Planetary Interiors
Van Nostrand Reinhold
New York

Indebetouw
R.
et al. 
2014
ApJ
782
L2

Ishigaki
M. N.
Tominaga
N.
Kobayashi
C.
Nomoto
K.
2014
ApJ
792
L32

Iwamoto
N.
Umeda
H.
Tominaga
N.
Nomoto
K.
Maeda
K.
2005
Science
309
451

Janson
M.
Bonavita
M.
Klahr
H.
Lafrenière
D.
Jayawardhana
R.
Zinnecker
H.
2011
ApJ
736
89

Ji
A. P.
Frebel
A.
Bromm
V.
2014
ApJ
782
95

Joggerst
C. C.
Woosley
S. E.
Heger
A.
2009
ApJ
693
1780

Johansen
A.
Youdin
A.
Mac Low
M.-M.
2009
ApJ
704
L75

Johnson
J. L.
Li
H.
2012
ApJ
751
81

Kane
S. R.
2007
MNRAS
380
1488

Keller
S. C.
et al. 
2014
Nature
506
463

Kobayashi
H.
Kimura
H.
Watanabe
S.-i.
Yamamoto
T.
Müller
S.
2011
Earth Planets Space
63
1067

Kornet
K.
Bodenheimer
P.
Różyczka
M.
Stepinski
T. F.
2005
A&A
430
1133

Kozasa
T.
Hasegawa
H.
Nomoto
K.
1989
ApJ
344
325

Kress
M. E.
Tielens
A. G. G. M.
2001
Meteorit. Planet. Sci.
36
75

Kuchner
M. J.
Seager
S.
2005
preprint (astro-ph/0504214)

Larimer
J. W.
1975
Geochim. Cosmochim. Acta
39
389

Léger
A.
et al. 
2004
Icarus
169
499

Lissauer
J. J.
1993
ARA&A
31
129

Llorca
J.
Casanova
I.
1998
Meteorit. Planet. Sci.
33

Loeb
A.
2014
Int. J. Astrobiology
13
337

Loeb
A.
Furlanetto
S. R.
2013
The First Galaxies in the Universe
Princeton Univ. Press
Princeton, NJ

Lucy
L. B.
Danziger
I. J.
Gouiffes
C.
Bouchet
P.
1989
Tenorio-Tagle
G.
Moles
M.
Melnick
J.
Lecture Notes in Physics, Vol. 350, Structure and Dynamics of the Interstellar Medium
Springer-Verlag
Berlin
164

Madhusudhan
N.
et al. 
2011
Nature
469
64

Madhusudhan
N.
Lee
K. K. M.
Mousis
O.
2012
ApJ
759
L40

Marassi
S.
Chiaki
G.
Schneider
R.
Limongi
M.
Omukai
K.
Nozawa
T.
Chieffi
A.
Yoshida
N.
2014
ApJ
794
100

Marassi
S.
Schneider
R.
Limongi
M.
Chieffi
A.
Bocchio
M.
Bianchi
S.
2015
MNRAS
454
4250

Mimura
K.
Sugisaki
R.
2003
Shock Reactions of Carbon-Bearing Materials and Their Cosmochemical Significance
Springer-Verlag
Berlin

Nagasawa
M.
Thommes
E. W.
Kenyon
S. J.
Bromley
B. C.
Lin
D. N. C.
2007
Reipurth
B.
Jewitt
D.
Keil
K.
Protostars and Planets V
Univ. Arizona Press
Tuscan, AZ
639

Naka
S.
Horii
K.
Takeda
Y.
Hanawa
T.
1976
Nature
259
38

Niedzielski
A.
Nowak
G.
Adamów
M.
Wolszczan
A.
2009
ApJ
707
768

Norris
J. E.
Ryan
S. G.
Beers
T. C.
1997
ApJ
489
L169

Norris
J. E.
et al. 
2013
ApJ
762
28

Nozawa
T.
Kozasa
T.
Umeda
H.
Maeda
K.
Nomoto
K.
2003
ApJ
598
785

Onodera
A.
Higashi
K.
Irie
Y.
1988
J. Mater. Sci.
23
422

Papaloizou
J. C. B.
Terquem
C.
2006
Rep. Prog. Phys.
69
119

Papoular
R.
Conard
J.
Guillois
O.
Nenner
I.
Reynaud
C.
Rouzaud
J.-N.
1996
A&A
315
222

Pinotti
R.
Arany-Prado
L.
Lyra
W.
Porto de Mello
G. F.
2005
MNRAS
364
29

Plez
B.
Cohen
J. G.
2005
A&A
434
1117

Poirier
J. P.
2000
Introduction to the Physics of the Earth's Interior
Cambridge Univ. Press
Cambridge

Rieke
G. H.
2008
Formation and evolution of terrestrial planets in protoplanetary and debris disks
Springer-Verlag
Berlin

Roederer
I. U.
Preston
G. W.
Thompson
I. B.
Shectman
S. A.
Sneden
C.
Burley
G. S.
Kelson
D. D.
2014
AJ
147
136

Salpeter
E. E.
Zapolsky
H. S.
1967
Phys. Rev.
158
876

Seager
S.
Mallén-Ornelas
G.
2003
ApJ
585
1038

Seager
S.
Kuchner
M.
Hier-Majumder
C. A.
Militzer
B.
2007
ApJ
669
1279

Takai
K.
Oga
M.
Sato
H.
Enoki
T.
Ohki
Y.
Taomoto
A.
Suenaga
K.
Iijima
S.
2003
Phys. Rev. B
67
214202

Tegmark
M.
Silk
J.
Rees
M. J.
Blanchard
A.
Abel
T.
Palla
F.
1997
ApJ
474
1

Tielens
A. G. G. M.
Seab
C. G.
Hollenbach
D. J.
McKee
C. F.
1987
ApJ
319
L109

Todini
P.
Ferrara
A.
2001
MNRAS
325
726

Tominaga
N.
Umeda
H.
Nomoto
K.
2007
ApJ
660
516

Tominaga
N.
Iwamoto
N.
Nomoto
K.
2014
ApJ
785
98

Umeda
H.
Nomoto
K.
2003
Nature
422
871

Umeda
H.
Nomoto
K.
2005
ApJ
619
427

Valencia
D.
O'Connell
R. J.
Sasselov
D.
2006
Icarus
181
545

Veras
D.
Crepp
J. R.
Ford
E. B.
2009
ApJ
696
1600

Vinet
P.
Ferrante
J.
Rose
J. H.
Smith
J. R.
1987
J. Geophys. Res.
92
9319

Vinet
P.
Rose
J. H.
Ferrante
J.
Smith
J. R.
1989
J. Phys.: Condens. Matter.
1
1941

Weidenschilling
S. J.
1988
Formation processes and time scales for meteorite parent bodies
Univ. of Arizona Press
Tuscan, AZ

Weidenschilling
S. J.
Cuzzi
J. N.
1993
Formation of Planetesimals in the Solar Nebula
Univ. of Arizona Press
Tuscan, AZ

Winn
J. N.
Fabrycky
D. C.
2015
ARA&A
53
409

Wisotzki
L.
Koehler
T.
Groote
D.
Reimers
D.
1996
A&AS
115
227

Yanny
B.
et al. 
2009
AJ
137
4377

Yasui
C.
Kobayashi
N.
Tokunaga
A. T.
Saito
M.
Tokoku
C.
2009
ApJ
705
54

Yong
D.
et al. 
2013
ApJ
762
27

York
D. G.
et al. 
2000
AJ
120
1579

Yoshida
N.
Abel
T.
Hernquist
L.
Sugiyama
N.
2003
ApJ
592
645

Zapolsky
H. S.
Salpeter
E. E.
1969
ApJ
158
809

Zinner
E.
1998
Meteorit. Planet. Sci.
33
549