AU Microscopii in the FUV: Observations in Quiescence, During Flares, and Implications for AU Mic b and c

Adina Feinstein, Kevin France, Allison Youngblood, Girish Duvvuri, DJ Teal, Wilson Cauley, Darryl Seligman, Eric Gaidos, Eliza Kempton, Jacob Bean, Hannah Diamond-Lowe, Elisabeth Newton, Sivan Ginzburg, Peter Plavchan, Peter Gao, & Hilke Schlichting


We observed AU Mic for 5 hours with HST/COS and detected 13 FUV flares.

We find strong continuum increases at λ ≤ 1120 Å during flares, which could be attributed to thermal bremsstrahlung processes.

We find no evidence of coronal dimming after or affiliated proton beams in our observations.

We evaluate the instantaneous atmospheric mass-loss for AU Mic b and find evidence that super-flares could result in detectable signatures of mass-loss.

Who is AU Mic?

AU Mic is a close (9.72pc) 22 Myr (Mamajek & Bell, 2014) M dwarf with 2 transiting exoplanets (Plavchan et al. 2020; Martioli et al. 2021; Gilbert et al. 2022). Given its age, this system is a crucial benchmark for understanding the role of stellar activity in shaping exoplanet atmospheres.
Understanding the contribution of stellar flares to atmospheric removal has only just begun. Garcia-Sage et al. (2017) modeled EUV-driven proton and O+ escape from an Earth-like planet around Proxima Centauri. They speculated that very energetic flares could increase the ionization fraction at low-altitudes, indirectly enhancing atmospheric escape.
Neves Ribeiro do Amaral et al. (2022) accounted for the X-ray and UV (XUV) contribution of flare flux in atmospheric escape from Earth-like planets around M dwarfs. The XUV flux from flares produced surface water loss for planets with mass Mp = 5 MEarth in their simulations. However, the effects of radiation from frequent highenergy flares on short period, young planets has not been fully investigated.


The FUV Time-Series and Spectra of Flares and in Quiescence


We observed AU Mic over 2 Hubble visits using the Cosmic Origins Spectrograph (COS) during transits of AU Mic b (PI Cauley) with the G130M grating from 1060-1360 Å. We masked Ly-ɑ to not damage the detector. We present the average quiescent spectrum to the left! The high SNR and high activity level of AU Mic produces a spectrum rich with emission features, which allow us to create a nearly complete list of present emission features, with respect to the databases and published FUV line lists. In total, we identified 176 emision features. We present a table of measured velocity shifts, fluxes, and FWHM for both quiescent and in-flares lines, which can be downloaded here.

We created 30-second light curves using the COS time-tag feature to study both the transit & flares. Due to the small data set, we identified flares by-eye in each orbit. To identify flares, we searched two separate light curves: the C III emission line at 1175.59 Å and the Si III emission line at 1294.55 Å (see Appendix Figure A1 in our paper). This method is consistent with previous studies (e.g. Woodgate et al. 1992) We identified 13 flares over our observations, yielding an FUV flare rate of 2.4 flares/hour! The flares in our sample have measured energies from 1x1030 to 2x1031 erg and equivalent durations ranging from 2 to 690 seconds.


Affiliated Physical Processes


Thermal Bremsstrahlung
Thermal bremsstrahlung is a principal emission mechanism for the Soft X-ray emission observed in solar flares (Shibata 1996; Warren et al. 2018; McTiernan et al. 2019). We fit for both the temperature and electron number density in a thermal bremsstrahlung profile at λ ≤ 1120 Å. We found temperatures of 9.1 ≤ log10(T) ≤ 11.2 best-fit the continua increases seen during Flares B, D, J, K, and M (see left). However, we note these fits converge only for electron number densities of ∼ 1022 cm−3, which is not representative of the stellar chromosphere. Therefore, thermal bremsstrahlung cannot be solely responsible for this feature, and it is unclear what other mechanisms may be contributing to this FUV excess.

Coronal Mass Ejections
One promising method to detect coronal mass ejections (CMEs) in time-series photometry is through the process typically referred to as “coronal dimming” (Harra et al. 2016; Veronig et al. 2021) which is caused by the depletion of plasma in the corona of a star during a CME (Hudson et al. 1996; Sterling & Hudson 1997). Recently, Veronig et al. (2021) reported three statistically significant (σ ≥ 4.4) dimming events in X-ray observations of AU Mic with depths ranging from 12 − 24%.
We searched for dimming events in our light curves created from the Fe XII, Fe XIX, and Fe XXI emission lines. These lines form at 106.2 K, 107.0 K, and 107.1 K, and trace the quiet and active corona, respectively. We searched for post-flare dimming during Flare D in these Fe lines. We find that the pre- and post-flare flux for all iron lines searched in this paper are within a 1σ agreement with each other, indicating no evidence of coronal dimming associated with Flare D.

Orrall-Zirker Effect
Low-energy (< 1 MeV) protons are challenging to detect because of the lack of affiliated X-ray or microwave radiation. However, it is possible that these protons could interact with and excite chromospheric hydrogen atoms. This process would subsequently result in spontaneous decay and the release of a high-energy photon. The high-energy photons are potentially detectable via flux excess in the red wing of Ly-α (Orrall & Zirker 1976).
To determine if there was an affiliated proton beam in our observations, we searched for enhancement in the blue and red wings of Ly-α, respectively. We created 1 second light curves from the third orbit of Visit 1 to search for an affiliated proton beam around both discrete peaks of Flare B from 1202 − 1204 Å and 1222 − 1227 Å. We found the overall detection of enhancement was non-significant. Therefore, we conclude that there is no evidence for the Orrall-Zirker Effect.

Implications for AU Mic b & c

What do these flares and newly measure FUV luminosity mean for AU Mic b? We evaluate photoevaporation-driven mass-loss with and without flares ( Owen & Wu, 2017; Owen & Campos Estrada, 2020). We adopt a quiescent luminosity LHE = 2.71 x 1029 erg s−1, which is calculated from our differential emission measurement (DEM) over 1 ≤ λ[Å] ≤ 1100, for which the DEM is reliable. We simulate the results for AU Mic b only, since AU Mic c does not satisfy the Jeans criteria. We evaluate the calculation over three scenarios: no flares (left), flares persistent for 200 Myr (middle), and flares persistent for 1 Gyr (right). The boxes are colored by the average mass-loss rate.


No Flares
We find that the median mass-loss rate for AU Mic b across all assumed core masses ranges from 1.6 to 2.5 x 108 g s−1. These calculations are consistent with the upper limit set by Hirano et al. (2020) using the metastable infrared He I triplet with NIRSPEC/Keck-II.

Flares for 200 Myr
When we include flares for the first 200 Myr, we find no significant change in the time-averaged mass loss rate, with minimal increases of up to 1.5x the no-flare baseline.

Flares for 1 Gyr
We find the time-averaged mass loss rate increases by 3x the no-flare baseline. We find super flares (L > 1034 erg s-1) can increase instantaneous mass-loss by 6 orders of magnitude, up to 1014 g s−1. This could lead to future variable transit depths/shapes.

For more details, be sure to check out our paper on the arXiv!