ALMA Discovery of Dust Belts around Proxima Centauri

The star (with a diameter of $\sim 0\buildrel{\prime\prime}\over{.} 001$) should appear angularly unresolved in the ALMA observations. Thus, the observed intensity peak in the ALMA 12 m array image (angular resolution $\simeq 0\buildrel{\prime\prime}\over{.} 7$) sets an upper limit of $100\pm 10\,\mu \mathrm{Jy}$ (Figure 2) for the stellar flux density.

The expected non-thermal emission of the star at 1.3 mm, extrapolated from the non-thermal flux density and spectral index measured by Slee et al. (2003), as well as our own ATCA measurements between 1 and 3 GHz (J. F. Gómez et al. 2017, in preparation), is negligible ($\ll 1\,\mu \mathrm{Jy}$).

The expected thermal emission from the star can be obtained from the photospheric emission model that fits the overall SED as described by Ribas et al. (2017), giving an extrapolated flux density at 1.3 mm of $74\pm 4\,\mu \mathrm{Jy}$. This result is fully consistent with the upper limit ($\lt 100\,\mu \mathrm{Jy}$) provided by our ALMA 12 m array observations (Figure 2), and indicates that ∼70%–80% of the flux density detected by ACA ($340\pm 60\,\mu \mathrm{Jy};$ Figure 1) does not arise from the star; therefore, we interpret it as thermal emission from circumstellar dust (see below).

As shown above, the flux density of the unresolved (diameter $\lesssim 6^{\prime\prime}$) source detected by ACA largely exceeds what can be accounted for by the star, and we interpret the excess of $\sim 270\,\mu \mathrm{Jy}$ as thermal emission from dust orbiting the star at radii $r\lesssim 3^{\prime\prime}$ ($r\lesssim 4$ au). The actual distribution of this dust can be further constrained by the ALMA 12 m array data.

If this emission were originated in a compact region (up to a few times the solid angle of the ALMA 12 m array synthesized beam), it would have been detected with a high S/N in the ALMA image (Figure 2). However, the image only reveals two sources above the $3\sigma$ threshold within a region of $\sim 6^{\prime\prime}$ in size (similar to the size of the ACA synthesized beam), totaling a flux density of $\sim 70\,\mu \mathrm{Jy}$ after the subtraction of the estimated emission of the star. This implies that the remaining $\sim 200\,\mu \mathrm{Jy}$ should be distributed over a solid angle $\gtrsim 7$ times that of the ALMA beam for its intensity to remain below the $3\sigma$ threshold of $30\,\mu \mathrm{Jy}$ beam⁻¹ (if a fraction of the emission was resolved out by the interferometer, this conclusion still holds). This condition requires that a significant part of the emission comes from radii larger than $\sim 1^{\prime\prime}$. Thus, with our current data we infer that there are ∼200 $\mu \mathrm{Jy}$ of dust emission spread over a belt in a range of radii from $\sim 1^{\prime\prime}$ to $\sim 3^{\prime\prime}$, corresponding to ∼1.3–4 au. A more precise determination of its distribution would require additional, more sensitive observations.

While our data are insufficient for a proper modeling, they are nevertheless enough to obtain a rough estimate of the masses involved. The mass of dust detected by ALMA can be estimated approximately as $({m}_{\mathrm{dust}}/{M}_{\oplus })=0.5$ ${({S}_{\nu }/\mathrm{Jy})({T}_{d}/{\rm{K}})}^{-1}$ ${(d/\mathrm{pc})}^{2}\,{(\nu /230\mathrm{GHz})}^{-2}$, where a dust opacity of ${\kappa }_{\nu }\,=2\,{\mathrm{cm}}^{2}\,{{\rm{g}}}^{-1}$ has been adopted (e.g., Beckwith et al. 2000). The dust temperature can be approximated as $({T}_{d}/{\rm{K}})\,={278(L/{L}_{\odot })}^{0.25}\,{(r/\mathrm{au})}^{-0.5}$ (Wyatt 2008). Thus, for the dust observed at r ≃ 1–4 au, we obtain ${T}_{d}\simeq 40$ K and a dust mass ${m}_{d}\simeq 4\times {10}^{-6}\,{M}_{\oplus }$.

The 1.3 mm continuum emission traces dust grains with μm to cm sizes. However, this population of small grains results from the collisional cascade involving a primordial population of larger bodies and planetesimals containing most of the mass. The equilibrium size distribution resulting from this collisional cascade can be described by a power-law of index −3.5 (Tanaka et al. 1996). Following a formulation similar to that of Wyatt & Dent (2002), it can easily be shown that, if the primordial size distribution connects smoothly with the collisional cascade distribution, the total mass can be approximated by ${m}_{\mathrm{tot}}\,\simeq {m}_{\mathrm{dust}}{({D}_{\max }/{D}_{\mathrm{dust}})}^{0.5}$, where ${D}_{\max }/{D}_{\mathrm{dust}}$ is the ratio between the maximum sizes of the population of large bodies and that of the observed dust emission. Taking ${D}_{\mathrm{dust}}\simeq 1\,\mathrm{cm}$, we obtain ${m}_{\mathrm{tot}}\simeq 2200\,{m}_{\mathrm{dust}}{({D}_{\max }/50\mathrm{km})}^{0.5}$. Therefore, if we integrate up to ${D}_{\max }\simeq 50$ km (e.g., Greaves et al. 2004; Wyatt et al. 2007b), we obtain a total mass of ${m}_{\mathrm{tot}}\simeq {10}^{-2}\,{M}_{\oplus }$.

This mass is similar to that of the solar Kuiper Belt ($\sim {10}^{-2}\,{M}_{\oplus };$ Bernstein et al. 2004), which also has a similar temperature (∼50 K), but is located at a much greater distance (30–50 au) from the Sun. Given the very low luminosity of the M-dwarf star Proxima Centauri (Section 1), one would expect that physical conditions similar to those required for the solar Kuiper Belt are attained at distances much closer to the star. Therefore, we suggest that the dust emission in Proxima Centauri, arising at scales ∼1–4 au, is likely tracing a Kuiper Belt analog around this star.

Additionally, we note that the central source detected by the ALMA 12 m array (Figure 2) appears marginally elongated along PA ≃ 130°, with a flux density of 106 $\mu \mathrm{Jy}$ and a deconvolved size of ∼0.8 au. A hint of an excess of emission in the proximity of the star and elongation along a similar PA is also observed in images made by combining the ACA and ALMA 12 m array data. Thus, considering the stellar emission to be 74 $\mu \mathrm{Jy}$ (see Section 3.1), it might be possible that a small amount of warmer dust, with a flux density of ∼30 $\mu \mathrm{Jy}$, is present at distances of ∼0.4 au from the star. Following the same procedures as above, we estimate a characteristic temperature ${T}_{d}\simeq 90$ K, a dust mass of $\sim 5.5\times {10}^{-7}\,{M}_{\oplus }$, and a total mass of $\sim {10}^{-3}\,{M}_{\oplus }$ for this possible hotter component.

The 1.3 mm ACA image (Figure 1) does not show direct evidence for dust structures other than the compact central source, but it shows a number of weak emission peaks that could be part of a larger structure. Since a tilted circular belt would appear as an ellipse on the sky, we performed averaging of the observed emission over elliptical annuli centered on the star in order to increase the S/N. When azimuthally averaging the intensity in annuli, the S/N of the intensity profile at a given radius improves by a factor equal to the square root of the ratio between the length of the annulus and the beam diameter. This approach has been successfully used to infer the presence of rings and gaps in protoplanetary disks around young stars (Osorio et al. 2014; Macías et al. 2017), as well as in debris disks around old stars (Marino et al. 2017).

The radial intensity profiles obtained in this way suggest an emission peak around a radius of $\sim 23^{\prime\prime}$ ($r\simeq 30$ au). We analyzed a grid of different position angles and eccentricities of the ellipses. The peak appears to be better defined and stronger for a combination of the ellipse PA ≃ 140° (major axis) and an eccentricity corresponding to an inclination angle $i\simeq 45^\circ$ (see Figures 3 and 1). Deviations from this combination of parameters produce a progressive vanishing of the feature. This feature is only detected in the ACA image because it appears at great distances from the center where the response of the ALMA 12 m primary beam is very low because of its smaller FWHM (see Section 2). If this belt proves to be real, it would provide a good estimate of the orientation of the orbital plane of the Proxima Centauri planetary system. If coplanarity is assumed, the true mass of the planet Proxima b (Anglada-Escudé et al. 2016) would be $\sim 1.8\,{M}_{\oplus }$. Interestingly, the PA of this outer belt is similar to that found for the elongated central source ($\sim 130^\circ ;$ Figure 2), supporting the reality of both structures and suggesting a similar orientation of the system at both small and large scales. However, we should emphasize that the detection of this outer belt is marginal, and it should be confirmed with additional observations of higher sensitivity. We note that the $45^\circ$ inclination of this tentative outer belt differs significantly from the $108^\circ$ inclination of the orbit of Proxima Centauri around Alpha Centauri AB (Kervella et al. 2017). However, we do not consider these results to be in conflict because, in general, orbital motions in a hierarchical triple system are not expected to be coplanar (e.g., Muñoz & Lai 2015). Indeed, the orbit of the Alpha Centauri AB binary also has a significantly different inclination ($79^\circ ;$ Kervella et al. 2016).

Zoom In Zoom Out Reset image size

Thus, it is possible that Proxima Centauri is surrounded by several belts of dust (see Figure 4), one or several close to the star ($r\lesssim 4$ au), and another one (to be confirmed) at a large radius ($r\simeq 30$ au). For this outer belt the average intensity is $\sim 76\,\mu \mathrm{Jy}$ beam⁻¹ (Figure 3), and we estimate a total flux density of ∼1.7 mJy. The presence of such a distant belt in the very low-luminosity star Proxima Centauri is challenging. It would be extremely cold (with ${T}_{d}\simeq 10$ K, if the same temperature law is assumed) and its flux density would lead to a dust mass of $1.4\times {10}^{-4}\,{M}_{\oplus }$, corresponding to a total mass (including large bodies) of $0.33\,{M}_{\oplus }$, if the same assumptions as in Section 3.2 are made, but see Krivov et al. (2013). This mass is much larger than that of the solar Kuiper Belt ($\sim {10}^{-2}\,{M}_{\oplus };$ Bernstein et al. 2004), and so far there is no known analog in the solar system. We note, however, that Herschel revealed a new class of very cold debris disks around some mature solar-type stars (Eiroa et al. 2011), the origin (still unexplained) of which could share some similarities with our proposed 30 au belt in Proxima Centauri.

Zoom In Zoom Out Reset image size

As noted in Section 2, our images show a marginally detected ($\sim 40\,\mu \mathrm{Jy}$ at $4\sigma$) compact source of 1.3 mm emission at a projected distance of $\sim 1\buildrel{\prime\prime}\over{.} 2$ southeast from the star (see Figure 2). This source is very intriguing. We can discard beam or cleaning artifacts, given the low beam sidelobes and high quality of the image. However, the significance of the source is marginal, and with the current data we cannot rule out the possibility that this component is just a noise peak.

If the source is real, then we can consider several possibilities. The source could be a background galaxy (Smail et al. 1997; Fujimoto et al. 2016). The probability of finding a source like this within $1\buildrel{\prime\prime}\over{.} 2$ of Proxima Centauri is small ($\lesssim$ 10⁻², according to the source counts of Fujimoto et al. 2016), but it cannot be completely discarded. Given the large proper motions of the star, a second-epoch observation would easily reveal whether this source moves together with the star or if it is a background static source. A substellar companion with a temperature of the order of 1000 K could also produce the observed emission. However, according to our calculations, such an object would produce a detectable signature in the radial velocity (RV) of the star that has not been observed. A transient event, such as the collision between large bodies (Wyatt et al. 2007a), might produce a cloud of dust with properties similar to the observed source. However, observation of such an episodic event seems unlikely in this old star. The source might be tracing a cloud of dust orbiting in the proximity of the Lagrange points of an still-undetected planet, in a way similar to the Trojan minor planets in our solar system. However, Trojan clouds are located at the L₄ and L₅ Lagrange points $\sim 60^\circ$ ahead and behind the larger body. If the $140^\circ$ position angle of the ∼30 au tilted belt is significant, and the inner dust disks and potential planetary orbits share the same inclination, their brightest parts would be approximately southeast and northwest, so we might be seeing the more favorably placed Trojan cloud, or possibly just the brightest part of an uneven disk that is mostly just below our detection threshold at the higher resolution.

Finally, an exciting alternative scenario is that the source traces a ring of dust surrounding an as-yet-undiscovered giant planet orbiting at a (projected) distance of 1.6 au (orbital period $\gtrsim 75.8$ year). By analogy with the rings of Saturn we expect a power-law distribution with an index of −3.5 and a maximum particle size of 5–10 m (Zebker et al. 1985; Brilliantov et al. 2015), resulting in a total mass of $\sim {10}^{-5}\,{M}_{\oplus }$ for such a planetary ring. Theoretical arguments (Charnoz et al. 2017) suggest that evolved planetary rings have a mass ∼10⁻⁷ times the mass of the planet. Thus, under this scenario, we would expect a planet of mass $\sim 100\,{M}_{\oplus }$, the mass of Saturn, to account for the observed 1.3 mm emission. There is no clear RV signal to indicate that such a planet is present in the data of the long-term monitoring of the star. Further observations are being undertaken to confirm or rule out this intriguing possibility. At any rate, our study shows that ALMA provides already the necessary sensitivity and resolution to detect rings around exoplanets in Alpha Centauri, and perhaps in other nearby stars.