JWST won’t find the very first stars, but a new telescope can

It was only three short years ago that we still lived in the Hubble era, with the most distant object then discovered being galaxy GN-z11, whose now-arriving light was emitted just 420 million years after the Big Bang. In July of 2022, however, JWST began science operations, and in just a few short months, there was a new cosmic record-holder, as its larger aperture, colder temperatures, and longer wavelength sensitivity allowed it to find objects that were invisible to a telescope like Hubble. As of the end of May, 2025, GN-z11 is now just the 14th most distant galaxy known, with the light from current record-holder MoM-z14 coming from just 280 million years after the Big Bang, or 140 million years earlier than GN-z11.

And yet, despite all our best endeavors to find them, any signature of the first stars remains elusive. We know that, immediately after the start of the hot Big Bang, there were no stars or galaxies; it takes at least many millions of years for gravity to pull matter together in sufficient densities to trigger the ignition of nuclear fusion. We’ve searched for signatures of these hypothetical population III stars, but no compelling ones have yet appeared. There are strong reasons to believe that these sought-after stellar signatures are beyond the capabilities of any existing telescope, even JWST. But there are two possible technological developments — and two novel telescope ideas — that could finally bring them within our reach. Here’s how we might actually find the very first stars of all.

Only because the most distant galaxy spotted by Hubble, GN-z11, is located in a region where the intergalactic medium is mostly reionized, was Hubble able to reveal it to us at the present time, breaking the prior record held by EGSY8p7. Other galaxies that are at this same distance but aren’t along a serendipitously greater-than-average line of sight as far as reionization goes can only be revealed at longer wavelengths, and by observatories such as JWST. At present, GN-z11 has been relegated to the 14th most distant galaxy known as of May 2025, nearly 3 years into the JWST era.

When Hubble found galaxy GN-z11, it was a bit of a surprise. After all, Hubble took its deepest-ever image by observing the same region of space for long periods of time across a wide variety of wavelengths: from the optical all the way to the longest infrared wavelengths that it’s sensitive to.

Remember: light from stars comes in a wide variety of wavelengths, but for the hottest, youngest, most massive stars, that radiation peaks in the ultraviolet part of the spectrum, with the strongest emission signatures coming from the Lyman-series emission lines of hydrogen. Early on, the Universe is also filled with neutral atoms, and those atoms absorb light of certain frequencies, including within the Lyman series as well; those atoms only become transparent to light (i.e., reionized) once a sufficiently large number of ultraviolet photons have been produced. Hubble was only able to detect GN-z11 because it happens to exist along a serendipitously more-reionized-than-average line of sight, and is right at the limit of its wavelength capabilities.

However, by virtue of it:

• being kept colder,
• being outfitted with instruments optimized for longer-wavelength viewing,
• and having a light-collecting area more than seven times as great as Hubble’s,

JWST has surpassed those limits. Instead of being limited to 400-500 million years after the Big Bang (or a redshift of ~10-11), JWST can take us back to fewer than 300 million years after the Big Bang, or a redshift of ~14+.

Before JWST, there were about 40 ultra-distant galaxy candidates known, primarily via Hubble’s observations. Early JWST results revealed many more ultra-distant galaxy candidates, but now a whopping 717 of them have been found in just the JADES 125 square-arcminute field-of-view. The entire night sky is more than 1 million times grander in scale. While some candidates will survive spectroscopic follow-up, others will not; no candidates at a redshift of 15 or above have survived spectroscopic follow up to date. Much science remains to be conducted, but very few facilities (other than JWST itself) are capable of conducting the needed follow-ups.

However, JWST still has limits of its own. JWST may be more powerful than Hubble in terms of light-gathering power, operating temperature, wavelengths sensitivity, and even instrument efficiency, but it’s still not all-powerful. In particular, it’s still limited by the fact that the earlier back in time we look — i.e., closer back in time towards the Big Bang — the greater the density of neutral atoms surrounding any point in space will be. This comes from two facts:

• the fact that the Universe is expanding, and so the matter density when the Universe was only a couple of hundred million years old is thousands of times greater than it is today,
• and the fact that once the Universe forms neutral atoms as it cools from its initially hot, dense state, those atoms remain neutral (and very good at absorbing ultraviolet and visible light) until a sufficient amount of stars form so that those atoms become reionized.

You might think that since JWST observes infrared light, it wouldn’t matter what happens to ultraviolet and visible light, but that’s incorrect. JWST detects these most-distant galaxies by observing light that’s in the infrared portion of the spectrum now, but back when those galaxies were first emitting their light, it was primarily ultraviolet light. As a result, JWST only detects the light that was:

• generated at short wavelengths,
• back when the Universe was filled with matter that absorbed light at those wavelengths,
• if there’s enough of that light (and little enough of that matter) to avoid being absorbed (or, as astronomers say, extincted) completely,
• so that it can arrive at our instruments today, after being stretched in wavelength by the expansion of the Universe.

This figure shows the NIRCam (top) and NIRSpec (bottom) data for now-confirmed galaxy MoM-z14: the most distant galaxy known to date as of May 2025. Completely invisible at wavelengths of 1.5 microns and below, its light is stretched by the expansion of the Universe. Emission features of various ionized atoms can be seen in the spectrum, below, as well as the significant and strong Lyman break feature.

Although JWST has taken us remarkably far back in cosmic history, all of the very high redshift (at z > 15 or so) galaxy candidates that it had identified have turned out to be impostors so far: intrinsically red later-time galaxies with strong emission lines, masquerading as ultra-distant galaxies but whose true nature is revealed by spectroscopy. In all the observations made so far, JWST has failed to take us back to the first ~250 million years of cosmic history, and in all the galaxies it has found, the material inside is well-evolved, showing strong evidence (including elemental signatures) that we aren’t seeing the first stars, but rather objects that have formed large populations of stars — including generations stars that have lived-and-died already — significantly prior to JWST’s observations.

And yet, the first stars must exist somewhere out there, beyond the limits of what we’ve found. Pristine populations of gas, containing hydrogen and helium alone, must have formed stars to lead to the presence of carbon, nitrogen, oxygen and more found in the galaxies that we do see. The observational challenge lies in discovering them. While the brute-force approach of gathering as much light as possible at the longest wavelengths possible over the longest time periods possible may yet reveal earlier, more distant galaxies than we’ve found so far, JWST is limited by the light-blocking material present at early times. If not enough light can get out of the galaxy and propagate through the neutral-matter-filled Universe without being fully extincted, JWST has no hope of detecting these still-earlier signatures.

The first stars and galaxies in the Universe will be surrounded by neutral atoms of (mostly) hydrogen gas, which absorbs the starlight. Without metals to cool them down or radiate energy away, only large-mass clumps in the heaviest-mass regions can form stars. The very first star will likely form at 50-to-100 million years of age, based on our best theories of structure formation and our best observations of the Universe to date, which corresponds to a redshift of between 30-and-50.

However, there is hope: if not from JWST, then from other observatories. Light (e.g., starlight) that’s emitted at short wavelengths, like ultraviolet and visible light, is easily absorbed by collections of neutral atoms, as shown above. But longer wavelength light, like infrared light and radio waves, can pass through those very same atoms virtually undisturbed. If there were some telltale signature of light that was:

• emitted at long wavelengths,
• including by stars or as a consequence of newly formed stars,
• that we could somehow detect even today despite their wavelengths being shifted by the expansion of the Universe,

then we could use that signature to try and find the very first stars, rather than being limited by the severe extinction of starlight at short wavelengths by the environment around those earliest stars.

While JWST can go partway into the mid-infrared part of the spectrum, out to wavelengths of around 28-30 microns at most, there are important astrophysical signatures generated at still-longer wavelengths. For example, there are very strong emission lines involving the elements carbon and oxygen in the far-infrared that ought to be generated as soon as the first stars form the first heavy elements (which are carbon and oxygen), with:

• singly-ionized carbon producing a line at 158 microns,
• neutral oxygen producing a line at 63 microns,
• and doubly ionized oxygen producing a line at 88 microns.

At extreme distances, these lines would be redshifted to millimeter (or even possibly centimeter) wavelengths, which a properly designed telescope could potentially observe.

One of doubly-ionized oxygen’s emission features peaks at 88 microns in the rest-frame: in the far-infrared. Owing to cosmic expansion, that light was stretched until it arrives at our eyes at ~millimeter wavelengths: in the right range for ALMA to be sensitive to it. For extreme redshift objects, greater sensitivity will be needed than even what ALMA can provide.

When stars are formed even in faint, nearby galaxies, these emission lines all appear, and so it stands to reason that they would be present in star-forming regions in the early Universe from the first moment that carbon and oxygen are present: once the very first stars have died and produced carbon and oxygen at all. It’s possible that with a powerful enough far-infrared (or even a near-radio) telescope, we could reveal the ultra-distant analogues of these nearby star-forming dwarf galaxies, enabling us to find evidence of stars far beyond whatever ultimate limits JWST will eventually run into. The flagship-class far-infrared space telescope concept, Origins, could serve as a template for the type of observatory we’d need for the task. (Probing imperfections in the cosmic infrared background, also suggested by some, is likely to be a less successful strategy.)

In other words, if you can find a signal that:

• is produced by stars or new star-formation,
• either directly or as a consequence/part of the aftermath of star-formation,
• that isn’t absorbed or extincted by the matter surrounding it,
• that can propagate freely through the Universe from its emission point until today,
• and that you can practically detect either from Earth, in space, or via another accessible location,

you’ll truly have a way to hunt down the very first stars of all, without being bound by the technical limitations of even a great observatory like JWST. While far-infrared lines offer a great possibility for detecting these first stars, there’s another possibility that might be even a more powerful probe: the 21 centimeter line produced by neutral hydrogen.

Whenever a neutral hydrogen atom forms, the electron within it will spontaneously de-excite until it’s in the lowest (1s) state of the atom. With a 50/50 chance of having those spins of the electron and proton aligned, half of those atoms will be able to quantum tunnel into the anti-aligned state, emitting radiation of 21 centimeters (1420 MHz) in the process. This should allow us to probe clumps of neutral hydrogen even farther back than the existence of the first stars.

Every time you produce or form a neutral hydrogen atom, it’s made out of two particles: the atomic nucleus, which is usually just a plain old proton, and a single electron that becomes bound to it. In general, we think of the hydrogen atom as a simple structure with energy levels, where the lowest energy level (or the ground state) of hydrogen is where all hydrogen atoms wind up without an external source of energy inputted into them.

However, electrons and protons both have a quantum mechanical property inherent to them known as spin: a measure of a particle’s internal angular momentum. Electrons and protons are both spin-½ particles, which means they can be “spin up” (with +½) or “spin down” (with -½), leading to two possibilities for ground-state hydrogen atoms. They can either have their spins anti-aligned, where one is spin up and the other is spin down, which is truly the lowest-energy state, or they can have their spins aligned, with both being spin up or both being spin down. There’s a 50/50 shot of having aligned or anti-aligned; it’s totally random each time you make a neutral hydrogen atom that reaches the ground state.

But that’s where the fun comes in: if you make a “spin aligned” hydrogen atom, it will eventually spontaneously flip in spin, emitting a line of a characteristic wavelength: 21.106114053 cm to be precise. This transition takes, on average, about 9 million years to occur, so a broad emission line should be expected to emerge from every cloud of such atoms that form each time they form.

This ground-based, wide-field image of the Eagle Nebula shows the star-forming region in all its glory, with new stars, the blue glow of reflected starlight, and the red glow of ionized atoms all present. When new stars form, hydrogen atoms get ionized, and as electrons recombine with protons, they emit light across a variety of wavelengths: ultraviolet, optical, and infrared. Once those atoms settle down into the ground state, however, their spins will flip about 50% of the time, creating 21 cm emission radiation.

So when and where do you form neutral hydrogen atoms? In a few places, with notable examples including:

• when the Universe becomes neutral for the first time, 380,000 years after the Big Bang (at a redshift of ~1089),
• when you inject a large amount of energy into a dense cloud of gas, such as from a burst of star-formation, and electrons and protons recombine to form hydrogen again,
• or in the environments found around active black holes, where injected energy can ionize atoms, where the protons and electrons then find each other once again to form neutral hydrogen.

When you transition from having an ionized plasma (with free protons and electrons) to having neutral atoms (i.e., protons and electrons bound together), the electrons will cascade down a variety of energy levels to make ground-state hydrogen. Then, over the next few several million years, 50% of these hydrogen atoms (the ones born with aligned spins) will spin-flip to emit 21 cm radiation, which then gets redshifted by the expansion of the Universe to even longer wavelengths. For various redshifts, this would produce light that was observed at a variety of wavelengths.

• At a redshift of z = 15 (age of 267 million years), the light would have a wavelength of 3.38 meters.
• At a redshift of z = 25 (age of 128 million years), the light would have a wavelength of 5.49 meters.
• At a redshift of z = 50 (age of 46 million years), the light would have a wavelength of 10.76 meters.
• And at a redshift of z = 1089 (age of 380,000 years), the light would have a wavelength of 230 meters.

On the left, the infrared light from the end of the Universe’s dark ages is shown, with the (foreground) stars subtracted out. 21 cm astronomy will be able to probe epochs in the Universe’s history even farther back than an observatory like JWST can directly search for signatures of the first stars, but sufficiently advanced observatories have not yet come around to rise to this challenge.

The wonderful thing about the 21 cm line is that there is no extinction for it: what gets produced travels freely throughout the Universe, unaffected by any neutral matter that intervenes along its travel path. Every time (and in every place) that you form neutral hydrogen atoms, this light gets produced, and if you can observe it, you can trace it back to its source of origin. Ideally, with a telescope (or array of telescopes) sensitive to these wavelengths of radiation, a long enough observing campaign could map out anywhere, at any time, that neutral atoms were newly produced, enabling to map out the history of star-formation in the Universe wherever we dare to survey it.

This requires radio observations, which isn’t so bad. These signals will be faint, but in principle, they are detectable. However, here on Earth, the background radio signals from our planet are unavoidable everywhere. Human activity produces far too many of them, both on Earth and in space above Earth, for even ALMA or the Event Horizon Telescope to detect these ultra-distant 21 cm sources. In fact, those signals appear all over the Solar System; we can’t hide from them anywhere.

Except, that is, in one place: on the far side of the Moon, or the portion that always faces away from us. It isn’t always dark, but it is always shielded from Earth, making it an ideal location to build a large radio telescope for exactly these purposes. Either a single-dish lunar crater radio telescope or a multi-dish array of far-side lunar radio telescopes could perfectly do the trick.

Constructing either a very large radio dish, perhaps in a lunar crater, or alternatively an array of radio telescopes, on the far side of the Moon, could enable unparalleled radio observations of the Universe, including in the all-important 21 centimeter range, both nearby and across cosmic time. The ability to map out where neutral hydrogen has newly formed within the past ~20 million years would advance our understanding of cosmic history like nothing else.

At the current moment in time, we’ve discovered objects back closer in time to the Big Bang than ever before. What JWST can do is impressive; it’s shattered a variety of cosmic records, including all of the top 10 (and more) spots for the most distant galaxy known. In time, with more observing time and more dedicated spectroscopic observations, it ought to break the current record for most distant galaxy several times over. The best from JWST is still yet to come.

But the fundamental limits of rest-frame extinction still exist, making it extremely unlikely that JWST will ever find a true population of pristine, first-generation, population III stars. To overcome those limits, we don’t need a bigger, more powerful JWST; we need telescopes optimized for longer-wavelength signatures: signatures that will be very faint and stretched-out in terms of redshift. It’s up to us how we choose to approach this problem, but the strongest options are going deeper into the infrared (such as the far-infrared, from space) or all the way into the radio (such as from the far side of the Moon), detecting signatures that are certain to emerge in the short-term aftermath of creating the very first stars.

The Universe, at some point, had to form stars for the very first time. Our ignorance about when that occurred — 200 million, 100 million, 50 million, or 30 million years after the Big Bang, for example — is staggering. And yet, at some point, if we look in the right way, we’re certain to find stars made of pristine material left over from the hot Big Bang. JWST can’t do it, but if we dare to explore the Universe in one of these two ways, the next generation of scientists almost certainly can.