The science of distant worlds is no longer just about whether a planet exists. It’s about what atmospheres cling to them, how winds gust inalienably at the edge of a star’s glare, and how the light that reaches us contains a memory of events that happened light-years away. In a recent study led by D. Mounzer of the University of Geneva and a large European collaboration, astronomers pushed the boundaries of how precisely we can read that memory. Using the ESPRESSO spectrograph on the Very Large Telescope, they hunted for fingerprints in a planet’s atmosphere so fine that they had to disentangle them from the star’s own flicker and from Earth’s atmospheric interference. The target was the archetypal hot Jupiter HD 189733 b, a world that has become a cornerstone in exoplanetary science because it orbits a bright, active star and transits with a rhythm that’s relatively easy to observe from the ground. The result is a richer, more nuanced narrative about what a planetary atmosphere can look like when the light from its star dances through it and then passes through our instruments on Earth.
The two-night campaign, conducted with ESPRESSO at an astonishing spectral resolution of about 140,000, let the team peer into the far edge of the line shapes that make up the sodium doublet in the planet’s atmosphere. This isn’t just a technical triumph. It’s a story about how measurement is limited not by our instruments alone but by our understanding of the star in front of the planet—and how that star can shape the very signals we’re trying to read as signatures of distant worlds. The researchers confronted a trio of challenges: planet-occulted line distortions that arise during transit because the star’s surface isn’t uniform, stellar activity that modulates the stellar spectrum on transit timescales, and subtle instrumental quirks that can masquerade as signals. The question they asked was simple and profound: can we separate the planet’s atmospheric fingerprints from the star’s own game of light, and if so, what does the planet’s atmosphere really look like? The answer, as you’ll see, is both sobering and exhilarating.
The ESPRESSO experiment and the stars’ whispers
Two transits of HD 189733 b were captured with the ESPRESSO spectrograph on the European Southern Observatory’s Very Large Telescope in Paranal, Chile. The team recorded spectra at the optical edge where the sodium doublet sits, achieving an effective resolving power of about 140 000. Each night produced more than 40 high-quality spectra, with exposures of 300 seconds. The observations were complemented by simultaneous photometric monitoring to track how the planet’s transit depth and the host star’s brightness behaved in real time, because the depth of the transit and the star’s limb darkening can subtly color the transmission spectrum.
To turn raw spectra into a usable transmission spectrum, the researchers used the antaress workflow, a careful chain of processing steps that includes correcting for Earth’s atmosphere, removing instrumental artifacts, and aligning all spectra to the star’s rest frame. They also wrestled with telluric lines from water vapor and oxygen, and with ESPRESSO’s own wiggles—a sinusoidal ripple imprinted by the telescope’s optics. These corrections aren’t cosmetic; they’re foundational. If you want to know what the planet’s atmosphere is doing, you first have to know what the star and the instrument aren’t doing, and you have to do it with precision that approaches the quantum of the data itself.
At the heart of the analysis is a family of models that describe how the star’s light would look if the planet weren’t there. The team used a technique called Rossiter-McLaughlin (RM) reconstruction to map how the star’s surface velocity field and line shapes change across the disk as the planet traverses it. The upshot is a constraint on the star’s rotation, differential rotation, convective blueshift, and even how the star’s spectrum changes from center to limb. With those stellar fingerprints in hand, the researchers could begin to isolate the planet’s absorption signal in the sodium doublet, a signature that has haunted exoplanet spectroscopy for more than a decade as being unsually tricky to interpret in the presence of POLDs—planet occulted line distortions—caused by RM and center-to-limb variation.
Disentangling the planet from the star
One of the paper’s crucial realizations is that even remarkably high-quality data can be dominated by POLDs if you don’t correct for them properly. The team explored several strategies to mitigate these distortions. They tried masking the stellar track around the sodium cores, effectively hiding the parts of the spectrum most heavily contaminated by the RM and center-to-limb effects. They also experimented with data-driven corrections that treat the master-out spectrum—the planet’s transit in a time-averaged frame—as a proxy for the local stellar spectrum occulted by the planet, but with a shift to the velocity corresponding to the occulted stellar region. Finally, they used synthetic stellar spectra generated with NLTE (non-local thermodynamic equilibrium) models, customizing the sodium abundance and the stellar temperature to match the observed star as closely as possible. Each approach dampened the POLDs in different ways, but none could erase them completely across the entire transit.
What the team could pin down with a high degree of confidence was a planetary signal that remains robust after careful POLD treatment, especially when the signal is viewed in the planet’s own rest frame and restricted to phases where the planet’s Doppler track and the star’s track don’t perfectly masquerade one another. The sodium absorption profile in the first half of the transit shows a clear, blueshifted planetary component, even after accounting for the star’s distortions. When the signals from the two lines of the doublet are co-added, the team measures a planetary absorption of 0.432 percent with a FWHM of about 3.82 kilometers per second, and a blueshift of roughly 7.97 kilometers per second toward the planet’s day-to-night terminator. That blueshift is the spectral imprint of winds aloft on the planet, pushing the atmosphere toward the observer on the near side of the terminator. It’s a striking, physically plausible signature of a dynamic atmosphere rather than a static veil of gas.
The analysis wasn’t just a single straight line fit. The researchers then engaged in a forward-modeling exercise with EvE, a code that simulates how stellar spectra would appear as the planet and its atmosphere obscure the star during transit. In EvE, one can tune the planet’s atmospheric temperature, sodium density, and even include a day-to-night wind to see how those choices reshape the observed absorption. The best-fitting planetary atmosphere, under a conservative one-dimensional hydrostatic framework, pointed to a temperature around 2750 K and a sodium density consistent with a tenuous, high-altitude layer at the Roche lobe. An added wind speed of about 8 km per second helped reconcile the observed blueshift with the model. In short: the planet is not a silent mask; it has winds that leave a measurable, Doppler-shifted signature in the light we detect on Earth.
But the study also laid bare the hard limits of the modeling. The authors found that synthetic 1D stellar spectra, even when carefully tuned, could not perfectly reproduce the actual POLDs, hinting that the Sun-like star HD 189733 hosted a star whose line cores and chromospheric contributions resist simple modeling. They experimented with artificially injecting chromospheric absorption into the stellar grid and with allowing the chromospheric component’s strength to vary during the transit. The result was a better qualitative match to the disc-integrated stellar spectrum and a more faithful representation of POLDs, but the residuals remained non-negligible. The takeaway is sobering: we are entering an era where the precision of our instruments is high enough that the star’s complexity—its activity, its outer atmospheric layers, its three-dimensional structure—becomes the dominant source of uncertainty in exoplanet atmospheres.
Sodium, lithium, and what they tell us
The two sodium lines in the HD 189733 b spectrum carry a twin message. First, after meticulous POLD correction, the planet shows a genuine sodium signature in its upper atmosphere. Second, the amplitude and shape of that signal depend on how you account for the star. In the first half of the transit, where the planet’s Doppler track and the star’s don’t violently collide, the team derived a robust planetary sodium absorption signal: 0.432 percent when the lines are co-added, with a width of about 3.8 kilometers per second and a blueshift near 8 kilometers per second. This is consistent with a strong day-to-night wind blowing gas along the terminator and with a moderately warm, extended sodium layer high in the atmosphere. The authors also performed a line-by-line fit to the individual D2 and D1 lines, obtaining blueshifts around 7.9 km/s for D2 and 4.8 km/s for D1, with amplitudes of roughly 0.38 and 0.46 percent, respectively. The consistency between lines strengthens the planetary interpretation, though the exact numbers carry the caveats of POLD residuals and the star’s own variability.
A surprising and striking find is the detection of lithium in the planetary atmosphere. The Li I line at 6707.775 angstroms rises clearly above the noise with an absorption of 0.102 percent and a blueshift of about 2.4 kilometers per second. That signal, at a significance of 6.4 sigma, is among the rare detections of lithium in exoplanet atmospheres and raises tantalizing questions about a planet’s primordial lithium reservoir or atmospheric processes that might preserve lithium against stellar destruction. The authors note that lithium sits outside the stellar spectrum as an easily isolated tracer, making it a particularly attractive probe for planetary atmospheres in high-precision data—though the line’s interpretation remains delicate, given the same star-planet interplay that complicates sodium.
Beyond sodium and lithium, the study surveyed other atomic species. Many showed POLD residuals or stellar chromospheric variability that overwhelmed any potential planetary signature. Magnesium, manganese, barium, and potassium either hovered at the noise or were tangled with telluric or chromospheric features. The absence of a clean, unambiguous signal for those species in this data set doesn’t mean they aren’t present; it underscores how challenging it is to separate planetary atmospheres from their stellar and terrestrial foregrounds at these exquisite levels of detail.
The stakes for the next generation of telescopes
This study isn’t merely a triumph of technique; it’s a reality check. The authors conclude that we have likely reached the practical limits of how accurately current one-dimensional stellar atmosphere models can reproduce the sodium line cores of a magnetically active K-type star during transit. The star’s chromosphere and the complex three-dimensional nature of stellar surfaces demand more sophisticated representations—likely three-dimensional magneto-hydrodynamic models—to parse the spectra cleanly. As ground-based spectrographs push toward the Extremely Large Telescope era, the same problems will intensify: higher signal-to-noise and higher spectral resolution will reveal even more subtle POLDs and stellar variability that could masquerade as atmospheric signatures if not modeled correctly.
In other words, the planet’s atmosphere can only be read as clearly as we understand the star it orbits. The authors advocate for investing in better stellar models and in multi-epoch monitoring that captures how a star’s spectrum stirs on timescales similar to a transit. They also highlight the promise of EvE-like forward modeling that simultaneously simulates the star, the planet, and their evolving interplay, rather than treating the stellar contribution as a static backdrop. The payoff is not only a more faithful inventory of elements in exoplanet atmospheres but also a deeper grasp of how atmospheric dynamics—winds, evaporation, condensation—play out in worlds far from our own.
From the author list’s breadth to the instrument’s precision, the HD 189733 b study is also a reminder of the collaborative nature of modern astronomy. The project brings together the University of Geneva’s Geneva Observatory as the leading institution, with researchers spanning institutes in Porto, Trieste, Bern, Montreal, Canaries, and beyond. The lead author D. Mounzer is a signal that a single person, anchored at a single institution, can catalyze a collaboration that makes us rethink how we observe planets around other stars—and how we interpret the faint whispers these worlds cast into our telescopes.
What this changes about our view of exoplanets
The message from this work is twofold. First, exoplanet atmospheres can show real, physically interpretable dynamical signatures even when the star and the instrument conspire to drown them in noise. The day-to-night winds inferred for the sodium layer of HD 189733 b give us a window into atmospheric circulation on a planet that’s heartbreakingly close to its star. It’s a portrait of a world where heat shuttles from the intensely irradiated day side to the cooler night side, carried by winds at thousands of kilometers per hour in a gas giant’s upper layer. Second, the study makes a strong case that the path to clean exoplanet spectroscopy runs through stellar physics. If we truly want to catalog atmospheres—water content, sodium abundance, metal signatures, and perhaps even biosignatures in the far future—we must first decode the star’s influence with the same care we apply to the planets themselves.
In the near term, the field will likely gravitate toward more sophisticated three-dimensional stellar models, time-resolved spectroscopy that maps how a star’s spectrum evolves during a transit, and multi-wavelength campaigns that separate telluric, stellar, and planetary components with cross-disciplinary methods. It’s a reminder that the frontier of exoplanet science is not only about building bigger telescopes but about building bigger, more nuanced models of the stars we watch through. The wind that blows across a distant planet’s terminator isn’t just a planetary phenomenon; it’s a signal filtered through a star’s atmosphere, a telescope’s optics, and our own need to understand both with humility and imagination.
As the community looks toward ELT-class facilities and even higher-resolution spectrographs, the lessons of this HD 189733 b study will guide how we design observations, process data, and interpret the light that travels across space and time. And it will keep reminding us that the best way to hear a world’s whisper is to listen closely to the star that lights it up—and to recognize that our tools, for all their power, are still learning to hear correctly in the cosmic orchestra.
Lead institutions: The study was led by the University of Geneva’s Geneva Observatory, with collaborators across Europe, including the Instituto de Astrofísica e Ciências do Espaço (Porto) and other European partners. The lead author is D. Mounzer.
