I’m enjoying the conversation about Project Hephaistos engendered by the article on Dyson spheres. In particular, Al Jackson and Alex Tolley have been kicking around the notion of Dyson sphere alternatives, ways of preserving a civilization that are, in Alex’s words, less ‘grabby’ and more accepting of their resource limitations. Or as Al puts it:

One would think that a civilization that can build a ‘Dyson Swarm’ for energy and natural resources would have a very advanced technology. Why then does that civilization not deploy an instrumentality more sly? Solving its energy needs in very subtle ways…

As pointed out in the article, a number of Dyson sphere searches have been mounted, but we are only now coming around to serious candidates, and at that only seven out of a vast search field. Two of these are shown in the figure below. We’re a long way from knowing what these infrared signatures actually represent, but let’s dig into the Project Hephaistos work from its latest paper in 2024 and also ponder what astronomers can do as they try to learn more.

Image: This is Figure 7 from the paper. Caption: SEDs [spectral energy distributions] of two Dyson spheres candidates and their photometric images. The SED panels include the model and data, with the dashed blue lines indicating the model without considering the emission in the infrared from the Dyson sphere and the solid black line indicating the model that includes the infrared flux from the Dyson sphere. Photometric images encompass one arcmin. All images are centered in the position of the candidates, according to Gaia DR3. All sources are clear mid-infrared emitters with no clear contaminators or signatures that indicate an obvious mid-infrared origin. The red circle marks the location of the star according to Gaia DR3. Credit: Suazo et al.

We need to consider just how much we can deduce from photometry. Measuring light from astronomical sources across different wavelengths is what photometry is about, allowing us to derive values of distance, temperature and composition. We’re also measuring the object’s luminosity, and this gets complicated in Dyson sphere terms. Just how does the photometry of a particular star change when a Dyson sphere either partially or completely encloses it? We saw previously that the latest paper from this ongoing search for evidence of astroengineering has developed its own models for this.

The model draws on earlier work from some of the co-authors of the paper we’re studying now. It relies on two approaches to the effect of a Dyson sphere on a star’s photometry. First, we need to model the obscuration of the star by the sphere itself. Beyond this, it’s essential to account for the re-emission of absorbed radiation at much longer wavelengths, as the megastructure – if we can call it that – gives off heat.

“[W]e model the stellar component as an obscured version of its original spectrum and the DS component as a blackbody whose brightness depends on the amount of radiation it collects,” write the authors of the 2022 paper I discussed in the last post. The modeling process is worth a post of its own, but instead I’ll send those interested to an even earlier work, a key 2014 paper from Jason Wright and colleagues, “The Ĝ Infrared Search for Extraterrestrial Civilizations with Large Energy Supplies. II. Framework, Strategy, and First Result.” The citation is at the end of the text.

The recently released 2024 paper from Hephaistos examined later data from Gaia (Data Release 3) while also incorporating the 2MASS and WISE photometry of some 5 million sources to create a list of stars that could potentially host a Dyson sphere. In the new paper, the authors home in on partial Dyson spheres, which will partially obscure the star’s light and would show varying effects depending on the level of completion. The waste heat generated in the mid-infrared would depend upon the degree to which the structure (or more likely, ‘swarm’) was completed as well as its effective temperature.

So we have a primary Dyson sphere signature in the form of excess heat, thermal emission that shows up at mid-infrared wavelengths, and that means we’re in an area of research that also involves other sources of such radiation. The dust in a circumstellar disk is one, heated by the light of the star and re-emitted at longer wavelengths. As we saw yesterday, all kinds of contamination are thus possible, but the data pipeline used by Project Hephaistos aims at screening out the great bulk of these.

Seven candidates for Dyson spheres survive the filter. All seven appear to be actual infrared sources that are free of contamination from dust or other sources. The researchers subjected the data to over 6 million models that took in 391 Dyson sphere effective temperatures. They modeled Dyson spheres in temperature ranges from 100 to 700 K, with covering factors (i.e., the extent of completion of the sphere) from 0.1 to 0.9. Among many factors considered here, they’re also wary of Hα (hydrogen alpha) emissions, which could flag the early stage of star growth and might be implicated in observations of infrared radiation.

Image: IC 2118, a giant cloud of gas and dust also known as the Witch Head Nebula. H-alpha emissions, which are observed over most of the Orion constellation, are shown in red. This H-alpha image was taken by the MDW Survey, a high-resolution astronomical survey of the entire night sky not affiliated with Project Hephaistos. I’m showing it to illustrate how pervasive and misleading Hα can be in a Dyson sphere search. Credit: Columbia University.

I want to be precise about what the authors are saying in this paper: “…we identified seven sources displaying mid-infrared flux excess of uncertain origin.” They are not, contra some sensational reports, saying they found Dyson spheres. These are candidates. But let’s dig in a bit, because the case is intriguing. From the paper:

Various processes involving circumstellar material surrounding a star, such as binary interactions, pre-main sequence stars, and warm debris disks, can contribute to the observed mid-infrared excess (e.g. Cotten & Song 2016). Kennedy & Wyatt (2013) estimates the occurrence rate of warm, bright dust. The occurrence rate is 1 over 100 for very young sources, whereas it becomes 1 over 10,000 for old systems (> 1 Gyr). However, the results of our variability check suggest that our sources are not young stars.

Are the candidate objects surrounded by warm debris disks? What’s interesting here is that all seven of these are M-class stars, and as the authors note, M-dwarf debris disks are quite rare, with only a few confirmed. Why this should be so is the object of continuing study, but both the temperature and luminosity of the candidate objects differs from typical debris disks. The questions deepen and multiply:

Extreme Debris Disks (EDD) (Balog et al. 2009), are examples of mid-infrared sources with high fractional luminosities (f > 0.01) that have higher temperatures compared to that of standard debris disks (Moór et al. 2021). Nevertheless, these sources have never been observed in connection with M dwarfs. Are our candidates’ strange young stars whose flux does not vary with time? Are these stars M-dwarf debris disks with an extreme fractional luminosity? Or something completely different?

The authors probe the possibilities. They consider chance alignments with distant infrared sources, and offsets in the astrometry when incorporating the WISE data. There is plenty to investigate here, and the paper suggests optical spectroscopy as a way of refuting false debris disks around M-dwarfs, which could help sort between the seven objects here identified. Stellar rotation, age and magnetic activity may also be factors that will need to be probed. But when all is said and done, we wind up with this:

…analyzing the spectral region around Hα can help us ultimately discard or verify the presence of young disks by analyzing the potential Hα emission. Spectroscopy in the MIR [mid-infrared] region would be very valuable when determining whether the emission corresponds to a single blackbody, as we assumed in our models. Additionally, spectroscopy can help us determine the real spectral type of our candidates and ultimately reject the presence of confounders.

So the hunt for Dyson spheres proceeds. Various pieces need to fall into place to make the case still more compelling, and we should remember that “The MIR data quality for these objects is typically quite low, and additional data is required to determine their nature.” This layman’s guess – and I am not qualified to do anything more than guess – is that rather than Dyson spheres we are glimpsing interesting astrophysics regarding M-dwarfs that this investigation will advance. In any case, do keep in mind that among some five million sources, only seven show compatibility with the Dyson sphere model.

If Dyson spheres are out there, they’re vanishingly rare. But finding just one would change everything.

The paper on Dyson sphere modeling is Wright et al., “The Ĝ Infrared Search for Extraterrestrial Civilizations with Large Energy Supplies. II. Framework, Strategy, and First Result,” The Astrophysical Journal Vol. 792, Issue 1 (September, 2014), id 27 (abstract). The 2022 paper from Project Hephaistos is Suazo et al., “Project Hephaistos – I. Upper limits on partial Dyson spheres in the Milky Way,” Vol. 512, Issue 2 (May 2022), 2988-3000 (abstract / preprint). The 2024 paper is Suazo et al., “Project Hephaistos – II. Dyson sphere candidates from Gaia DR3, 2MASS, and WISE,” MNRAS (6 May 2024), stae1186 (abstract / preprint).