PEPSI Investigation, Retrieval, and Atlas of Numerous Giant Atmospheres (PIRANGA). II. Phase-Resolved Cross-Correlation Transmission Spectroscopy of KELT-20b

TL;DR

This work applies high-resolution cross-correlation transmission spectroscopy to a single KELT-20b transit, yielding robust detections of Fe I ($11.9\sigma$) and Fe II ($23.7\sigma$) and tentative detections of Na I and Cr I. Phase-resolved analysis across eight orbital phase bins reveals distinct dynamical regimes: Fe II shows a strong, phase-dependent blueshift with limb asymmetry, while Fe I exhibits weaker, more modest blueshifts, indicating day-to-night winds and potential scale-height-driven limb effects. The study emphasizes reproducibility by outlining explicit detection criteria and demonstrates that correcting for systemic velocity differences reduces cross-study scatter, though some discrepancies persist between different instruments and analyses. The results broadly align with dynamical models that include drag and magnetic effects, while also highlighting tensions with some prior observations and underscoring the need for coordinated multi-instrument campaigns to fully map KELT-20b’s multidimensional atmosphere.

Abstract

KELT-20b is a well-studied ($T_{\text{eq}}=2262$ K) ultra hot Jupiter, but its multidimensional atmospheric structure remains unconstrained. We performed high-resolution cross-correlation transmission spectroscopy (HRCCTS) on a single transit time series of KELT-20b, observed with PEPSI on the LBT. Upon combining nineteen in-transit exposures, we detect Fe I $(11.9σ)$ and Fe II $(23.7σ)$ and tentatively detect Na I $(3.4σ)$ and Cr I $(3.3σ)$. The full-transit velocity offsets of the strongest absorbers are $ΔV_{\text{Fe I}} = -1.0 \pm 0.7$ km s$^{-1}$ and $ΔV_{\text{Fe II}}= 0.0\pm 0.5$ km s$^{-1}$, which are mostly inconsistent with previously published values for KELT-20b, although the previous measurements are mostly inconsistent with each other. By correcting for discrepant systemic velocity solutions of up to $1.7$ km s$^{-1}$ between studies, our Fe II offset becomes consistent with previous measurements ($\leq 1.7σ$), while Fe I remains significantly less blueshifted than in earlier studies ($ \geq 2.2-4.5σ$). We propose a set of detection criteria to improve future reproducibility in HRCCTS work. Phase-resolving the Fe I and Fe II absorption signatures into eight orbital phase bins reveals distinct dynamical regimes: Fe II exhibits a strong phase-dependent blueshift from ingress to egress along with significant limb asymmetry, while Fe I shows weaker signals and a more modest blueshift with phase. These patterns indicate day-to-night winds and suggest scale height differences are a significant driver of limb asymmetry in KELT-20b.

Figures (15)

• Figure 1: S/N, seeing, and airmass versus the corresponding orbital phase at the center of each exposure. The red and blue arms of PEPSI are color-coded. For the Seeing value, we use SEEINGSX, derived from measurements on the left side of the LBT. The vertical lines correspond to $T_1$, $T_2$, $T_3$, $T_4$, in order. There is a marked increase in S/N with phase, which impacts our interpretation of time-resolved CCFs, further discussed in Section \ref{['subsec:Atmospheric Dynamics']}.
• Figure 2: Removal of the Doppler shadow in the blue arm 2D CCF of Fe II. Top: 2D CCF of the blue arm Fe II signal for all $N_{\text{spec}}$ exposures stacked. The Doppler Shadow (blue streak) intersects the atmospheric velocity signal (yellow streak near $\mathrm{RV}=0$, at the center of the Doppler shadow). Middle: Doppler shadow CCF generated by our model and fitted to our CCF data via non-linear least squares optimization. Bottom: Residual CCF after subtracting off the Doppler shadow.
• Figure 3: Combined arm 2D full-transit CCFs of species exceeding the tentative detection threshold. Top left: Fe I. Top right: Fe II. Bottom left: Na I. Bottom right: Cr I.
• Figure 4: Left: Scatter plot of the Fe I and Fe II full-transit binned velocity offsets in studies that detect both species. We find the lowest wind speeds for each species, but the blueshift of Fe I relative to Fe II is somewhat preserved: Fe I is blueshifted relative to Fe II by $\sim-1$ km s$^{-1}$, which is inconsistent with other studies. Fe II is consistent with zero, in agreement with CARMENES observations by Nugroho2020 and roughly in agreement with Hoeijmakers2020. Our findings are inconsistent with the remaining studies of KELT-20b which detected both Fe I and Fe II. Right: The same plot but with systemic velocity discrepancies $\Delta v_{\text{sys}}$ corrected for. Our Fe II velocity offset is consistent with all Nugroho2020 observations. In the legend, $\Delta{\rm Fe\,I}$ and $\Delta{\rm Fe\,II}$ give, for each study, the difference between that study's Fe I/Fe II velocity offsets and ours, expressed in units of the combined $1\sigma$ uncertainty. The quoted $\tilde{\chi}^{2}$ is the reduced chi-squared obtained by summing the squared tensions in Fe I and Fe II and dividing by the number of species.
• Figure 5: Top: Coadded Fe I 1D CCFs from $T_{1C}$ in green, $T_{C4}$ in blue, and full-transit in black. The gray horizontal dashed line is the tentative detection threshold and the black horizontal dashed line is the detection threshold. The center of a Gaussian fit to the $T_{1C}$ and $T_{C4}$ 1D CCFs are displayed in the legend. Bottom: The same plot for Fe II.