Have we just found the first stars in the Universe?

We’ve only ever seen 2nd-generation stars and later. Until, just maybe, now.

“For my part I know nothing with any certainty, but the sight of the stars makes me dream.” –Vincent Van Gogh

Yet when you dream of the stars — of every star every eye has ever laid its gaze upon — these are polluted stars. As pure and pristine as the brilliant Sun seems, its continuous shining belies a dark and catastrophic history.

You see, it’s true that our Sun fuses hydrogen into helium at its core: the two lightest elements in the Universe, both formed in great abundance in the Big Bang, with helium also formed in prior generations of stars. But just around 2% of the elements in the Sun are heavier than the two lightest ones, made up of:

• 1% oxygen,
• 0.4% carbon,
• 0.14% iron,
• 0.1% silicon,
• 0.1% nitrogen,
• 0.08% magnesium,
• 0.06% neon,
• 0.04% sulphur,

and many others. All told, some sixty-five additional elements have been detected in the Sun.

Yet… the first stars weren’t like this at all! The Sun formed only after the Universe was already more than nine billion years old, meaning that countless generations of stars had lived, burned through their fuel, created heavy elements, and died, recycling their remnants back into the Universe. These enriched materials gave rise to subsequent generations of stars, complete with heavy elements, complex molecules, worlds with rocky cores, icy or watery surfaces, and — to put it mildly — so much more.

As time goes on and more generations of stars live, die, and are reborn from this process of cosmic recycling, the complexity of what the Universe can form skyrockets.

But immediately following the Big Bang, the Universe consisted of nothing more than hydrogen, helium, and less than 0.0000001% of anything heavier than that.

As our Universe expands and cools, we not only form these atomic nuclei, but then neutral atoms, and then the densest clumps of gas begin to contract under their own gravity. By time the Universe is 50-to-100 million years old, the very first stars begin to form, with this pristine hydrogen-and-helium igniting nuclear fusion in the cores of the first stars, producing helium and, in short order, a whole lot more.

But these stars don’t remain pristine for long! By time just half-a-billion years have gone by, more than 99% of the neutral atoms that we’ve formed have been reionized by the hot, ultraviolet starlight eminating from these behemoths, the first galaxies have already formed, and in the densest places, many generations of stars have lived-and-died.

The Universe is well underway by time our most powerful space telescopes can catch a glimpse of what it actually looks like.

But a few of these places in space, even hundreds of millions or even maybe a billion years after the Big Bang, are still likely pristine, unaffected by star formation. If you can find an isolated region of space, one that has never been polluted by these prior generations of stars, you can find the first pristine sample of the leftover atoms from the Universe’s birth. Although this might seem like a crazy idea, a team found not just one but two molecular clouds just like this back in 2011, finding isolated clouds with hydrogen, helium, deuterium, and nothing else.

But this is just neutral gas. Eventually, if this neutral gas gets dense enough, it should form the most elusive of all classes of stars: Population III stars, which to date, have only been theorized. Unlike super metal-rich stars like our Sun (Population I), which have seen many, many generations of stars form before, or metal-poor (Population II) stars, found in the halos of galaxies and in very young galaxies, where only a few generations of stars have lived-and-died before, these stars should be formed out of the gas left over from the Big Bang and nothing else.

Population III stars have only been theorized up until now… but with new observations, we may have just changed all of that.

A team led by David Sobral has just spectroscopically confirmed that one of the most distant, luminous galaxies in the Universe may be housing a group of Population III stars, which would make it the first observational signature for this set of stars that must have existed, but that we’ve never seen before.

Yet at the same time, we have to be careful when it comes to these stars; we could quite easily fool ourselves, and be mistaking a near-pristine sample for a truly pristine one. Our observations have limits, after all, and this is uncharted territory for science. So rather than write all about it for you from my non-expert point of view (I may be an astrophysicist, but I’m a theoretical cosmologist, not a star-formation specialist), I decided to turn to the star formation specialist himself, David Sobral, first author on this find! He was kind enough to grant me an exclusive interview, and what follows is my questions — and his responses — in full. (My questions are in bold; his responses are in italics.)

Let’s give you an artist’s impression video of this cluster, known as COSMOS Redshift 7 (or CR7, for you Cristiano Ronaldo fans), and dive right in!

Population III stars are supposed to be the first stars in the Universe, formed from the elements left over from Big Bang Nucleosynthesis and nothing more. What leads us to suspect that these stars are still pristine?

The spectrum of the source, all the way from the optical to the near-infrared (observed) reveals a very bright Lyman-alpha emission line, a very strong HeII emission line (basically helium is completely ionised) at rest-frame 1640Å, but, surprisingly, no other emission lines. Because we see HeII 1640Å, it means the source is extremely hot and highly energetic in the UV, and thus we would expect to see e.g. CIII], CIV], OIII], etc emission lines. These are seen in every single source with strong HeII emission, even in the most metal poor sources known.

[Ethan’s note: numbers like I, II, III, etc., after an element indicate its ionization state, corresponding to 0, 1, 2, etc. ionized electrons. The transition down to that state in the atom is denoted CIII if the transition is not forbidden, CIII] if it’s semi-forbidden, or [CIII] if it is forbidden, quantum mechanically.]

However, in the case of CR7, all other metal lines remain undetected, and we already have quite strong line limits e.g. HeII/OIII] 1663Å >3 and HeII/CIII] 1908Å > 2.5. Even the most metal poor galaxies have line ratios of typically unity and lower, so this at least places the galaxy on the most metal-poor we have ever seen. We are aiming to use HST and the grism to push these line ratios/limits even further and really test the metal-free scenario. Another key thing about the finding is that we expect that two components within the galaxy should already be enriched and contain more “normal” stellar populations: deep HST grism observations may allow us to detect heavy elements from such enriched components — if this were to happen it would further strengthen our interpretation of a “PopIII wave” moving away from the initial site of PopIII star-formation.

The completion of reionization, where enough stars have formed to ionize the intergalactic medium and make the Universe transparent to visible light, happens about 550 million years after the Big Bang, with stars having existed for hundreds of millions of years before that. Why do we suspect there will still be pristine material left this late in the game?

It is obviously hard to define “completion” of reionisation, but we know that it certainly is not 100% even after 800 million years. We know that e.g. the Lyman-alpha Luminosity function and the lyman-alpha fraction (along with studies from quasars) that there is still some neutral hydrogen above z~6. That said, we also know that most of reionisation happen prior to that (however, the most recent CMB results keep pushing the “instantaneous” moment for reionisation to later and later times).

However, our findings are not inconsistent at all with this. Actually, we *need* previous generations of stars to cave a big enough bubble around this galaxy (at least ~1 Mpc) for us to be able to see such bright Lyman-alpha line (otherwise it would simply get absorbed before redshifting out of the rest-frame Lyman-alpha, and would not reach us). Basically, a single PopIII episode of star-formation will never be seen, before reionisation is complete, in Lyman-alpha: by the time the UV photons have caved a big enough bubble, the most massive stars will be dead, and no Lyman-alpha will be produced.

So what we are actually seeing is something different, and consistent with some theoretical predictions of a “PopIII wave”. This happens in simulations, where some galaxies, even down to redshifts of ~3 or so, manage to keep some pockets of pristine gas, and thus PopIII SF gets delayed. This can happen for a number of reasons. In the case of CR7, the two “clumps” where most of the stellar mass is (redder) seem to have been more than capable of preventing PopIII star-formation a few kpc away due to the intense Lyman Werner radiation. So while such UV photons produced by previous generations of stars in the evolved parts contributed towards ionising a bubble, and preventing the formation of PopIII stars around, they likely only contaminated the 1–2kpc around them, while light reached further out.

Previous studies have found pristine gas even later than this, but have come from absorption lines illuminated by background quasars. How do the techniques of emission lines allow us to conclude that this gas is pristine, simply from the absence of certain excited elements?

As far as I know, no “pristine” gas was ever found. Very metal poor, yes, but I’m not aware of any study that found true “pristine gas”. The point in this case is the strong HeII 1640 Å line. If you completely ionise Helium and produce this emission line, you should also produce strong CIII] and OIII] emission lines. And we don’t see them.

[Ethan’s note: David has clarified that he refers to “pristine gas” within the halos of galaxies only when he claims that no “pristine” gas has ever been found. This is not to suggest that the samples found by Michele Fumagalli, John M. O’Meara, and J. Xavier Prochaska, referred to earlier, is polluted!]

There is a redder (i.e., evolved) population of stars very close to this suspected Pop III sample. What is the risk of contamination, here, and why would we still expect this young, blue cluster to be pristine?

The redder clumps are absolutely essential for the interpretation: not only to explain why we can see Lyman-alpha, but also to explain why this potential “last wave of PopIII star formation” is possible. We need the other clumps to have produced lots of UV photons in the past, and we know they are now likely enriched. From simulations, we also know that light travels much quicker and efficiently than metals.

To what level can we rule out the presence of carbon-and-oxygen (the first heavy elements formed in stars) in these stars based on the data available, and to what level would we need to rule them out to be convinced we’ve got a true Pop III sample here?

With new data from HST/Grism (WFC3) we will get to limits where the discussion between “ultra-metal-poor” and metal free becomes pointless. On the other hand, obtaining a JWST spectrum in the MIR will leave no doubts at all: if this is really metal free, then we will only see H-beta and various helium lines. If there are at least some metals, we will see a strong [OIII] emission.

[Ethan’s note: this line is totally forbidden, since it’s the 5007 Å line only seen in very low-density regions.]

Is there anything about Population III stars and the James Webb Space Telescope that you think the world should know?

JWST will not only allow us to really understand the nature of this source, but may even show us whether this is much more common than thought. In principle, JWST could reveal strong Helium lines in the MIR when starting to take spectra of the most distant galaxies we know about today. That would likely show that some PopIII star-formation goes on for longer than expected, and will also make us revise our models and interpretation of high-redshift sources.

And finally, is there anything else, either about this sample or about the detection of a pristine sample of material that you’d like to share?

One thing that makes this source really exciting, apart from being ultra-bright in Lyman-alpha and Helium II, but showing no metal lines, is the interpretation. The PopIII wave scenario may work, but the alternative would be equally exciting: a direct collapse black hole. Basically both need metal free gas, and both need a previous significant population of stars to delay the formation of stars while not enriching it. CR7 seems to be in the ideal conditions for both.

Thanks to David for providing all this excellent detail on what may turn out to be the very first true Population III stars in the Universe, something I was convinced we’d have to wait until James Webb to find. The really nice thing about this find is that someday soon, we can put it into context. Sobral’s team, using the most advanced telescopes in the world, are expanding their spectroscopic searches for these types of galaxies, at the highest redshifts, both closer and farther than CR7.

While Hubble may or may not find us another sample like this, JWST should give us plenty of Population III stars, and then we can truly find out whether CR7 is not only a part of that sample, but whether it might, in fact, be the last, latest-time Population III stars present in our entire observable Universe!

What a moment to be alive.

Leave your comments at our forum, and support Starts With A Bang on Patreon!