Confirmation of Fe I on MASCARA-5 b's Dayside Observed With EXPRES

Abstract: MASCARA-5~b/TOI-1431~b is an ultra-hot Jupiter \citep[$P_{\rm orb}=2.650237\pm0.000003\,{\rm d}$, $T_{\rm eq}=2370\pm70\,{\rm K}$, $M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$, $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$;][]{addison2021} orbiting a bright Am star ($V=8.0\,{\rm mag}$). Recent time-series observations obtained with PEPSI@LBT during the planet's post-eclipse phases have revealed Fe~{\sc i} emission lines indicative of a thermally inverted atmosphere. These observations demonstrate that MASCARA-5~b is well-suited to atmospheric characterization via emission spectroscopy, thereby motivating further follow-up observations covering additional orbital phases to constrain the planet's atmospheric chemistry, thermal structure, and dynamics. Here we present pre-eclipse time-series observations obtained with the high-resolution optical spectrograph EXPRES@LDT. Our analysis confirms the previous detection of gas-phase Fe~{\sc i} on MASCARA-5~b's dayside (with a $5.5σ$ significance obtained from two nights of observations) and the fact that the thermal profile is inverted with lower and upper temperatures $\sim2000\,{\rm K}$ and $\sim4500\,{\rm K}$, respectively. A search for Fe~{\sc ii} and Cr~{\sc i} did not yield any plausible detections. We also find that the pre-eclipse signal exhibits a non-negligible blueshift of $-3.2\pm1.4\,{\rm km/s}$ potentially caused by winds.

• The paper confirms gas-phase Fe I emission from MASCARA-5 b’s dayside at 5.5σ using high-resolution EXPRES spectra from two pre-eclipse nights.
• Cross-correlation and retrieval analyses measured Kp = 187.1 ± 3.4 km/s and a blueshift of −3.2 ± 1.4 km/s, consistent with planetary winds or other atmospheric dynamics.
• The results support a thermally inverted atmosphere spanning roughly 2000–4500 K, while non-detections of Fe II and Cr I leave metallicity, thermal structure, and wind mechanisms unresolved.

This paper by Sikora et al. presents pre-eclipse, high-resolution optical time-series observations of the ultra-hot Jupiter (UHJ) MASCARA-5 b/TOI-1431 b obtained with EXPRES on the 4.3 m Lowell Discovery Telescope. Using high-resolution cross-correlation (HRCC) techniques, the authors confirm the presence of gas-phase Fe I in emission on the planet's dayside at $5.5\sigma$ significance from two nights of data, corroborating an earlier PEPSI@LBT detection (2605.00154). The analysis also confirms a thermally inverted atmosphere with lower and upper temperatures of roughly 2000 K and 4500 K, and reports a non-negligible blueshift of $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$ that is plausibly attributable to atmospheric winds.

MASCARA-5 b is a UHJ with $P_{\rm orb}=2.650237\pm0.000003$ d, $T_{\rm eq}=2370\pm70$ K, $M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$, and $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$ orbiting a bright ($V=8.0$ mag) Am star. TESS photometry yields dayside and nightside temperatures of $3004\pm64$ K and $2583\pm63$ K, implying efficient day–night heat transport. Prior EXPRES Rossiter–McLaughlin measurements established a misaligned orbit ($\lambda=-155^{+20}_{-10}$ degrees), while earlier transmission searches found no atmospheric absorption, consistent with the planet's high surface gravity and small scale height. Petz et al. subsequently reported a $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$0 Fe I emission detection and tentative Cr I detection using post-eclipse PEPSI data; their pre-eclipse PEPSI sequence did not yield robust detections, attributed to lower S/N ($\Delta v = -3.2 \pm 1.4\,{\rm km/s}$1 versus $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$2) and fewer exposures.

The new observations comprise four pre-eclipse nights (phases $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$3–0.48) between October 2023 and June 2025: two higher-S/N nights with 46 exposures each of 350 s, and two lower-S/N nights with 35 and 31 exposures, likely affected by poor or variable seeing. EXPRES provides $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$4 over 3800–8220 Å, with ThAr wavelength solutions preferred over LFC for their wider coverage. Data were reduced with the EXPRES pipeline, tellurics modeled with SELENITE, and barycentric corrections applied pixel-by-pixel via the chromatic exposure meter.

The cleaning pipeline masks order edges, six heavily telluric-contaminated orders, and columns with $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$5 outliers; spectra are shifted to the stellar rest frame using RVs derived from A0-tuned CCFs (median RV errors of 2.5–4 m/s). After median normalization, three iterations of SYSREM remove most stellar residuals; additional sigma clipping and a 151-pixel median box filter complete the detrending. The authors tested alternatives—more SYSREM iterations, polyfit detrending, alternative masking—and found no improvement; detection significance for nights 1+2 rises from $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$6 at one SYSREM iteration to a maximum of $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$7 at three iterations, remaining at $\Delta v = -3.2 \pm 1.4\,{\rm km/s}$8–$\Delta v = -3.2 \pm 1.4\,{\rm km/s}$9 up to ten iterations.

Model emission templates for Fe I, Fe II, and Cr I were computed with petitRADTRANS v3 at $P_{\rm orb}=2.650237\pm0.000003$0, adopting a three-layer PT profile approximating that of the PEPSI study ($P_{\rm orb}=2.650237\pm0.000003$1 K at $P_{\rm orb}=2.650237\pm0.000003$2 bar; $P_{\rm orb}=2.650237\pm0.000003$3 K at $P_{\rm orb}=2.650237\pm0.000003$4 bar), FastChem equilibrium abundances at stellar metallicity $P_{\rm orb}=2.650237\pm0.000003$5, instrumental broadening to $P_{\rm orb}=2.650237\pm0.000003$6, and rotational broadening of $P_{\rm orb}=2.650237\pm0.000003$7 assuming tidal locking. The Gibson et al. matrix transformation accounts for SYSREM distortion of the templates. Cross-correlations were integrated over $P_{\rm orb}=2.650237\pm0.000003$8–$P_{\rm orb}=2.650237\pm0.000003$9 grids spanning $T_{\rm eq}=2370\pm70$0 and $T_{\rm eq}=2370\pm70$1 respectively.

Only the first two, higher-S/N nights show peaks consistent with the expected planetary track: combining them yields a $T_{\rm eq}=2370\pm70$2 Fe I detection at $T_{\rm eq}=2370\pm70$3 and $T_{\rm eq}=2370\pm70$4. The last two nights show no plausible planetary peaks ($T_{\rm eq}=2370\pm70$5), and neither Fe II nor Cr I is detected in any combination. Masking $T_{\rm eq}=2370\pm70$6 to suppress residual stellar lines raises the nights-1+2 significance to $T_{\rm eq}=2370\pm70$7 while all other cases remain $T_{\rm eq}=2370\pm70$8; notably, combining all four nights lowers the detection significance regardless of masking, which the authors attribute to the lower-quality later data diluting the signal.

An MCMC retrieval over three free parameters ($T_{\rm eq}=2370\pm70$9, $M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$0, and template scaling $M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$1) yields:

Parameter	Value
$M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$2	$M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$3
$M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$4	$M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$5
$M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$6	$M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$7

The recovered $M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$8 agrees closely with the value of $M_{\rm p}=3.12\pm0.18\,M_{\rm Jup}$9 calculated from published system parameters, supporting the planetary origin of the signal. The scaling parameter implies observed Fe I line strengths about 50% weaker than the nominal model, which could be produced by lowering the upper-atmosphere temperature from 4750 K to roughly 4000 K, or equivalently by a shallower temperature gradient or reduced Fe abundance. Re-running the injection-recovery MCMC with Gibson-style uncertainties shifts $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$0 only marginally to $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$1, indicating limited bias from uncertainty treatment.

Two independent validations strengthen the result. First, a Welch's $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$2-test comparing in-trail and out-of-trail CCF distributions (means of 0.655 and $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$3) rejects the null hypothesis at $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$4. Second, injection-recovery tests with signals injected at negative $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$5 recover the injected Fe I template at $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$6 when scaled by $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$7, with retrieved parameters discrepant from the injected values by only $R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$8 ($R_{\rm p}=1.49\pm0.05\,R_{\rm Jup}$9), $V=8.0$0 ($V=8.0$1), and $V=8.0$2 ($V=8.0$3); all non-detection cases show spurious peaks of $V=8.0$4–$V=8.0$5, establishing the effective detection threshold.

The measured blueshift of $V=8.0$6 is comparable to offsets reported for other UHJs, including the ESPRESSO detection of blueshifted Fe I emission on WASP-76 b's dayside ($V=8.0$7 overall, with $V=8.0$8 pre-eclipse versus $V=8.0$9 post-eclipse). Although the PEPSI study does not report a $3004\pm64$0 for its post-eclipse MASCARA-5 b detection, its published map suggests $3004\pm64$1, hinting at a possible decrease in blueshift from pre- to post-eclipse phases analogous to WASP-76 b. Candidate explanations include day-to-night winds, planetary rotation, magnetic drag, or biases from neglecting 3D atmospheric structure in the 1D templates—the latter being most significant near quadrature, though these observations lie closer to eclipse where such effects are smaller. Distinguishing among these mechanisms requires higher-S/N coverage of both pre- and post-eclipse phases.

The paper concedes several constraints on its conclusions. No full atmospheric retrieval was performed because only Fe I was detected; initial retrievals showed strong degeneracies among PT parameters, metallicity, and Fe abundance, so the thermal structure quoted ($3004\pm64$2 K, $3004\pm64$3 K) rests on fixing all parameters except velocity and line-strength scaling to the PEPSI-based model—an approximation further imperfect because the PEPSI Guillot-profile cannot be exactly replicated by the adopted Brogi-style three-layer profile. The non-detections of Fe II and Cr I are not interpreted as abundance constraints but simply as below-threshold signals, with spurious peaks of up to $3004\pm64$4 defining the noise floor. The physical origin of the blueshift remains unresolved, and the suggested phase dependence of $3004\pm64$5 is based on visual inspection of the PEPSI map rather than a quantitative measurement. Open questions left by the paper include whether higher-S/N optical observations can detect Fe II (which would break metallicity–thermal-structure degeneracies), whether near-IR CO, H$3004\pm64$6O, and OH can be detected with instruments such as IGRINS-2, and what mechanism produces the inferred blueshift and any phase dependence thereof.

Sikora et al. provide an independent confirmation of thermally inverted, Fe I-emitting dayside atmosphere of MASCARA-5 b using EXPRES, achieving a $3004\pm64$7 detection from two pre-eclipse nights validated by Welch's $3004\pm64$8-test and injection-recovery tests, with a recovered $3004\pm64$9 matching the dynamical expectation and a statistically significant $2583\pm63$0 blueshift. The work also marks the first successful measurement of a planetary dayside emission spectrum with EXPRES, demonstrating that spectrographs on modest-sized telescopes can contribute to UHJ emission studies. The remaining uncertainties—single-species retrievability, the wind-versus-rotation-versus-magnetic-drag origin of the blueshift, and the absence of Fe II and Cr I detections—define the specific observational targets for future follow-up of this system.