🔭 astrophysics

The HST/WFC3 Transmission Spectrum of AU Mic b Part I: An Atmosphere Obscured by Contamination and Systematics

This study presents the HST/WFC3 transmission spectrum of the young sub-Neptune AU Mic b, revealing that the data is heavily compromised by instrumental scanning instability and the star's intense magnetic activity, which obscures atmospheric features and limits constraints to a preference for a small atmospheric scale height dominated by the transit light source effect.

Original authors: William C. Waalkes, Peter Gao, Elisabeth Newton, Brett M. Morris, Zachory K. Berta-Thompson, Hannah R. Wakeford, Lili Alderson, Andrew W. Mann, Peter Plavchan, Patrick J. Lowrance, Natasha E. Batalha

Published 2026-07-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: William C. Waalkes, Peter Gao, Elisabeth Newton, Brett M. Morris, Zachory K. Berta-Thompson, Hannah R. Wakeford, Lili Alderson, Andrew W. Mann, Peter Plavchan, Patrick J. Lowrance, Natasha E. Batalha, Eric D. Lopez, Roxana Lupu

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Young Planet with a "Dirty" View

Imagine trying to take a clear photo of a tiny, colorful butterfly (the planet AU Mic b) flying in front of a very bright, flickering, and stormy lighthouse (the star AU Mic).

This paper is about a team of astronomers who tried to take a "transit spectrum" of that butterfly. A transit spectrum is like looking at the butterfly's shadow to see what gases are in its atmosphere. However, the team found that the view was so obscured by the lighthouse's flickering and the telescope's own wobbles that they couldn't clearly see the butterfly's colors. Instead, they mostly saw the lighthouse's glare.

The Cast of Characters

AU Mic b: A young, "sub-Neptune" sized planet (about 4 times the size of Earth) that is only about 20 million years old. It's like a teenager in the cosmic sense—still growing and holding onto a thick, primordial atmosphere.
AU Mic (The Star): A young, active M-dwarf star. Think of it as a hyperactive toddler who is constantly throwing tantrums (flares) and has huge, dark freckles (starspots) all over its face.
Hubble Space Telescope (HST): The camera used to take the picture. Unfortunately, during these specific observations, the camera was shaking.

The Two Main Problems

The researchers faced two massive hurdles that made their job incredibly difficult:

1. The Shaky Camera (Instrumental Systematics)
Imagine trying to read a book while someone is shaking the table you are sitting at. The letters (wavelengths of light) would wobble in and out of your vision.

What happened: Hubble experienced an "unstable scan." The telescope's pointing was shaky, causing the image of the star to jitter across the detector.
The result: This created a "variable PSF" (Point Spread Function), which is a fancy way of saying the light from the star smeared and shifted unpredictably. This made the baseline of the data (the starting point of the measurement) jump up and down from orbit to orbit.
The fix: The team tried to "bin" the data (grouping the shaky pixels together into larger buckets) to smooth out the wobble. It helped a little, but they couldn't fix it completely. This limited how precise their final results could be.

2. The Stormy Star (Stellar Contamination)
Imagine trying to listen to a whisper (the planet's atmosphere) while a loud, crackling fire (the star's activity) is right next to you.

What happened: The star AU Mic is covered in "spots" (cooler, darker patches) and "faculae" (hotter, brighter patches). As the star spins, these spots rotate in and out of view.
The "Transit Light Source" (TLS) Effect: When the planet passes in front of the star, it doesn't just block a uniform light. If the planet passes in front of a dark spot, it blocks less light than if it passes over a bright spot. This tricks the telescope into thinking the planet is a different size or has a different atmosphere than it actually does.
The result: The star's "freckles" left a massive imprint on the data, completely overwhelming the faint signal of the planet's atmosphere.

What They Found

Despite the noise, the team managed to learn a few things:

Mapping the Star: By analyzing the light when the planet wasn't in front of the star, they figured out the star's surface temperature. They found the star is about 3,891 K (Kelvin) on average, but covered in spots that are about 3,020 K. Roughly 33% of the star's surface is covered in these dark spots.
The Planet's Atmosphere: They tried to use computer models to guess what the planet's atmosphere looked like. They tested three scenarios:
1. Full Model: Atmosphere + Star spots.
2. TLS-Only: Just Star spots (no atmosphere features).
3. Step Function: A flat line with no features at all.
The Verdict: The data strongly rejected the "flat line" (meaning there is some signal), but it could not tell the difference between the "Full Model" and the "TLS-Only" model.
- Translation: The signal from the star's spots was so loud that it drowned out the planet's atmosphere. The data fits just as well if the planet has a thick, clear atmosphere as it does if the planet has a flat, featureless atmosphere. The star's "noise" is the dominant feature.

The Scale Height Surprise

One specific measurement they could make was the "scale height" (how "puffy" the atmosphere is).

Other young planets around normal stars have very "puffy" atmospheres (scale heights around 1,000 km).
AU Mic b's atmosphere appears to be much "flatter" or "denser," with a scale height of less than 185 km.
Why? This suggests AU Mic b is either much heavier (more gravity pulling the atmosphere down), has a very heavy atmosphere (high metallicity), or is covered in thick clouds/hazes that hide the lower layers.

What's Next?

The paper concludes that looking at this specific planet with Hubble was like trying to hear a whisper in a hurricane. The "TLS effect" (the star's spots) completely dominated the view.

The authors predict that even future observations with the James Webb Space Telescope (JWST) will likely be dominated by this same star-spot effect. They suggest that to truly understand this planet, scientists might need to look at the planet's emission (the heat it glows with) rather than its transit (the shadow it casts), as the glow is less affected by the star's spots.

In short: The team successfully measured the star's spots and confirmed the planet exists, but the "static" from the star was too loud to hear the planet's atmospheric "voice."

Technical Summary: The HST/WFC3 Transmission Spectrum of AU Mic b Part I

Problem Statement
Young sub-Neptune progenitors orbiting M dwarfs offer a critical window into planet formation and early evolution, potentially revealing the pristine conditions before significant atmospheric loss occurs. AU Mic b, a $\sim$ 20 Myr old sub-Neptune orbiting a nearby (9.7 pc) pre-main sequence M0Ve star, is an ideal candidate for such study. However, characterizing the atmospheres of planets around young, active M dwarfs is complicated by intense stellar variability. Specifically, the "Transit Light Source" (TLS) effect—where unocculted starspots and faculae alter the apparent transit depth—can mimic or obscure atmospheric features. Furthermore, the Hubble Space Telescope (HST) observations of AU Mic b were plagued by an unstable telescope scan, resulting in a variable Point Spread Function (PSF) that introduced significant orbit-to-orbit systematics. The primary challenge addressed in this work is disentangling the planetary transmission spectrum from these combined instrumental and stellar contaminants to determine if atmospheric features can be constrained.

Methodology
The authors analyzed two HST/WFC3 transits of AU Mic b observed in August 2021 (G141 grism, 1.14–1.64 $\mu$ m) and April 2022 (G102 grism, 0.79–1.13 $\mu$ m). The data reduction and analysis followed a multi-stage approach:

Data Reduction and Binning: Using the PACMAN pipeline, the authors reduced the raw data. To mitigate the variable PSF caused by unstable scanning, they employed a custom wavelength binning scheme. Instead of uniform bins, they placed bin edges based on the residual RMS of linear fits to out-of-transit (OOT) light curves and the normalized flux offset between orbits. This reduced spectral resolution but minimized the impact of the variable PSF on the baseline.
Stellar Characterization (OOT SED): The authors modeled the out-of-transit stellar spectral energy distribution (SED) using a multi-temperature spectral decomposition model (BT-Settl AGSS2009 library). They tested 1, 2, and 3-temperature models to constrain the photospheric temperature ( $T_{phot}$ ), spot temperature ( $T_{spot}$ ), and spot filling factor ( $f_{spot}$ ).
Light Curve Modeling: White light curves were modeled using two approaches: a "spotted" model (using the fleck code to account for rotational modulation and spot occultations) and a linear baseline model. The spotted model identified specific spot crossings on the transit chord, which were then trimmed from the spectroscopic light curves. Instrumental systematics (ramp and breathing effects) were modeled using a combination of exponential ramps and 4th-order polynomials in HST orbital phase.
Atmospheric Retrieval: Bayesian retrievals were performed on the resulting transmission spectrum using the PLATON code. Three models were compared:
- Full Model: Includes both atmospheric parameters (metallicity, C/O, temperature, clouds, hazes) and the TLS effect (parameterized by $T_{phot}$ , $T_{spot}$ , and $f_{spot}$ ).
- TLS-Only: Assumes a flat atmospheric spectrum with all features attributed to the TLS effect.
- Step Function: A featureless spectrum with only a dataset offset.

Key Results

Stellar Parameters: The 3-temperature model was strongly preferred for the stellar SED. The authors constrained the photospheric temperature to $3891 \pm 37$ K and the spot temperature to $3020 \pm 69$ K, with a spot filling factor of $0.33 \pm 0.05$ . A third, hotter component ( $\sim$ 6000 K) was detected but likely compensates for unresolved systematics rather than representing a physical hot spot.
Instrumental Systematics: The unstable scan created a variable PSF that manifested as orbit-to-orbit baseline shifts. While binning mitigated this, residual effects remained the primary limitation on precision.
Transmission Spectrum: The retrieved spectrum is dominated by the TLS effect. The data strongly prefers the "Full Model" and "TLS-Only" cases over a featureless "Step Function" (decisive evidence), but the data cannot distinguish between the "Full Model" and "TLS-Only" cases. The retrieved atmospheric parameters (mass, metallicity, temperature) are poorly constrained.
Atmospheric Scale Height: Despite the weak constraints, the retrieval places a strict upper limit on the atmospheric scale height of $<185$ km (at $3\sigma$ ). This is approximately 1/5th the scale height of young sub-Neptunes orbiting G dwarfs (e.g., V1298 Tau b, HIP 67522 b).
Consistency: The spot filling factor derived from the atmospheric retrieval ( $0.36 \pm 0.06$ for the Full Model) is consistent with the value derived independently from the OOT SED ( $0.33 \pm 0.05$ ), suggesting the TLS modeling is self-consistent.

Significance and Claims
The paper concludes that the transmission spectrum of AU Mic b is currently dominated by stellar contamination (the TLS effect) and HST systematics, rendering the atmospheric features weakly constrained. The authors modestly claim that the data is insufficient to detect molecular features like H $_2$ O or CH $_4$ with high confidence.

The significance of this work lies in:

Demonstrating the TLS Dominance: It provides a clear case study where the TLS effect overwhelms atmospheric signals in a young M-dwarf system, cautioning against over-interpreting spectral slopes or features in similar systems without robust stellar modeling.
Characterizing AU Mic b's Nature: The derived upper limit on the scale height suggests AU Mic b is distinct from low-density, clear-atmosphere young planets around G stars. It likely possesses a higher mass, higher metallicity, or a thick cloud/haze layer, implying different formation or evolutionary pathways for sub-Neptunes around M dwarfs.
Methodological Framework: The paper establishes a rigorous framework for handling unstable HST scans and stellar activity, setting the stage for "Paper II," which will present an empirical model to decontaminate the spectrum.
Predictions for JWST: The authors predict that JWST observations (NIRCam F322W2 and F444W) will likely also be dominated by the TLS effect, with only minor atmospheric contributions, and that optical transit depths could be $\sim$ 25% higher than measured in the near-IR due to the TLS effect.

The work underscores the challenges in characterizing young, active planetary systems and motivates the development of better empirical models for stellar photospheres to unlock the atmospheric properties of these primordial worlds.