Thermal Eclipse Observation of the Young Hot Neptune AU Mic b with Spitzer

We present the observation of a secondary eclipse of the young hot Neptune, AU Mic b, in the infrared using the Spitzer Space Telescope. Using a primary transit from Spitzer to constrain the system parameters, we tentatively detect an eclipse centered at $BJD=2458740.848893^{+0.00010}_{-0.000099}$ with an observed depth of $171\pm{29}$ ppm given an uninformed prior. This corresponds to a dayside brightness temperature of $T=1031\pm{58}$ K, which exceeds the calculated equilibrium temperature of $606\pm{19}$ K. We explore some possible explanations for these results, including inefficient heat redistribution, gravitational contraction, stellar pulsations, instrument systematics and choice of eclipse depth prior, but find none of these to be likely explanations for the observed eclipse parameters. We also explore the impact of correlated noise in the systematic trends, and we find that splitting the systematics into low-pass (smoothing) and high-pass trends is required to reach an optimal minimization of the low-frequency systematics in the resulting detrended light curve. Future observations with JWST are needed to confirm our eclipse detection with Spitzer.

💡 Research Summary

The authors present a search for the thermal secondary eclipse of the young hot Neptune AU Mic b using the Spitzer Space Telescope’s IRAC 4.5 µm channel. AU Mic is a 22 Myr, M1V star at ~9.7 pc that hosts a resonant chain of three transiting planets; AU Mic b has a mass of ~20 M⊕, radius ~4.2 R⊕, and an orbital period of 8.463 days. The Spitzer observations were obtained on 2019 Sept 14, immediately following a primary transit observed on Sept 10. The data consist of an 8‑hour continuous stare in sub‑array mode (0.08 s exposures, 0.1 s cadence) with the target placed on the well‑characterized “sweet‑spot” pixel.

The reduction follows the standard Demory et al. (2012) pipeline: conversion to photon counts, aperture photometry with radii from 2.0 to 4.0 pixels, sky annulus subtraction, and iterative sigma‑clipping of flux, background, centroid positions, and FWHM. The optimal aperture (3.2 pixels) is chosen by minimizing the out‑of‑eclipse RMS. Systematics inherent to Spitzer—primarily intra‑pixel sensitivity variations and pointing jitter—are modeled by separating low‑frequency trends (e.g., thermal drifts, background changes) from high‑frequency trends (pointing jitter) using a low‑pass/high‑pass decomposition. This dual‑trend approach reduces the residual RMS by ~15 % compared with a single polynomial detrending and better isolates the eclipse signal.

A Bayesian analysis with EXOFASTv2 is performed using two priors on the eclipse depth: an uninformed (effectively flat) prior and a uniform prior bounded between 0 and 500 ppm. Both yield a consistent eclipse depth of 171 ± 29 ppm, corresponding to a 5‑σ detection. The time of mid‑eclipse is BJD 2458740.848893 (+0.00010/‑0.000099), in agreement with the ephemeris derived from the primary transit. Residual noise analysis (time‑averaging RMS, β‑factor) indicates modest correlated noise (β ≈ 1.2–1.4), but not enough to invalidate the detection.

Converting the measured depth to a brightness temperature assuming blackbody emission gives T_day = 1031 ± 58 K. This is substantially higher than the equilibrium temperature calculated from stellar irradiation (T_eq ≈ 606 ± 19 K) assuming full redistribution and zero albedo. The authors explore several explanations: (i) inefficient heat redistribution (dayside‑only emission), (ii) residual internal heat from gravitational contraction, (iii) stellar variability (flares, spots) contaminating the infrared flux, (iv) instrument systematics not fully captured by the detrending, and (v) prior‑induced bias. They use TESS and Evryscope light curves to characterize the star’s rotational modulation and flare activity during the Spitzer window, finding that the star’s infrared brightness was near its median level and that flares were rare, making stellar contamination unlikely. Simulated injections of systematic trends show that the low‑pass/high‑pass model adequately recovers the eclipse depth, reducing the chance that the signal is an artifact of over‑fitting.

Despite the careful analysis, the authors caution that Spitzer’s limited photometric precision and the presence of low‑level correlated noise prevent a definitive claim. They advocate for follow‑up observations with the James Webb Space Telescope (JWST) using MIRI or NIRCam, which can achieve sub‑10 ppm precision and provide spectrally resolved eclipse measurements. Such data would confirm the depth, constrain the planet’s emission spectrum, and discriminate between a high‑day‑side temperature caused by poor heat redistribution versus an additional internal heat source.

In summary, this work reports a tentative 5‑σ detection of a 171 ppm secondary eclipse of AU Mic b, implying a dayside temperature far above equilibrium. The analysis introduces a novel low‑pass/high‑pass detrending scheme to handle Spitzer systematics and demonstrates that, even for a young, active M dwarf host, a thermal eclipse can be measured with Spitzer. Confirmation with JWST will be essential to solidify these findings and to advance our understanding of the thermal evolution of young Neptune‑mass exoplanets.