HAT-P-70b: Ultra-hot Jupiter Dynamics

• HAT-P-70b is an ultra-hot Jupiter orbiting a rapidly rotating A-type star with a 2.74-day orbit and an equilibrium temperature near 2560 K.
• High-resolution spectroscopy reveals a chemically rich atmosphere dominated by both neutral and ionized metals, with robust detections of Fe, Ca, and Ti.
• Advanced retrieval methods highlight key challenges including nickel enhancement, residual titanium depletion, and signatures of hydrodynamic escape.

Searching arXiv for recent HAT-P-70b papers to ground the article in the current literature. HAT-P-70b is an ultra-hot Jupiter orbiting the rapidly rotating A-type star HAT-P-70, also identified as HD 287325, on a short-period orbit of approximately 2.744 days. With an equilibrium temperature of $2562^{+43}_{-52}\,\mathrm{K}$, a radius of $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$, and a projected spin-orbit angle of $107.9^{+2.0}_{-1.7}{}^\circ$, it is both strongly irradiated and highly misaligned. High-resolution spectroscopy has established HAT-P-70b as one of the chemically richest ultra-hot Jupiter atmospheres yet characterized, with transmission and dayside-emission studies revealing a dense inventory of neutral and ionized metals, strong dynamical signatures, evidence for a thermally inverted dayside, and modeling sensitivities tied to ionization, condensation, and non-equilibrium upper-atmosphere physics (Sun et al., 8 May 2026, Bello-Arufe et al., 2021, Guo et al., 25 Dec 2025, Gan et al., 7 Dec 2025).

1. System architecture and irradiation regime

HAT-P-70b orbits an A-type, very rapid rotator with $T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$, $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$, $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$, and $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$. The host has $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$, making line broadening and Doppler-shadow modeling central to spectroscopic analysis. The planet’s orbit has $P = 2.744320 \pm 0.000001$ d in the MAROON-X study and $P = 2.74432452^{+0.00000079}_{-0.00000068}$ d in later CARMENES/PEPSI analyses, with semi-major axis $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$0 in the latter compilation. The radial-velocity upper limit is $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$1 at $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$2, while spectroscopic analyses converge on a mass near $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$3: $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$4 from HARPS-N and $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$5 from the MAROON-X equilibrium-chemistry retrieval, albeit with a bimodal posterior (Sun et al., 8 May 2026, Bello-Arufe et al., 2021).

The equilibrium temperature is quoted as

$1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$6

with $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$7, giving $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$8. This places HAT-P-70b in the “titanium transition” regime: hotter than WASP-76b and WASP-121b, but cooler than WASP-189b or KELT-9b, such that titanium is expected largely in the gas phase while nightside condensation may still matter. The planet is regarded as very likely tidally locked, implying persistent dayside forcing, cooler nightside conditions, and a transmission geometry sensitive to day-night advection, ionization gradients, and partial cold trapping (Sun et al., 8 May 2026).

The orbital geometry is unusually extreme. Doppler-shadow modeling of HARPS-N data refined the projected obliquity to $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$9, with $107.9^{+2.0}_{-1.7}{}^\circ$0, $107.9^{+2.0}_{-1.7}{}^\circ$1, and $107.9^{+2.0}_{-1.7}{}^\circ$2. This places the system in a nearly polar, highly misaligned configuration, consistent with the broader trend of hot Jupiters around hot stars exhibiting large obliquities (Bello-Arufe et al., 2021).

2. Observational access and high-resolution methodologies

The atmosphere of HAT-P-70b has been characterized with several high-resolution spectrographs. HARPS-N observed a single transit at $107.9^{+2.0}_{-1.7}{}^\circ$3 over $107.9^{+2.0}_{-1.7}{}^\circ$4–$107.9^{+2.0}_{-1.7}{}^\circ$5 Å, providing the first transmission-spectrum inventory (Bello-Arufe et al., 2021). MAROON-X on Gemini North later observed two full transits at $107.9^{+2.0}_{-1.7}{}^\circ$6, with wavelength coverage of 500–670 nm in the blue arm and 650–920 nm in the red arm; each transit consisted of 36 exposures, with 9 out-of-transit exposures and mean S/N per pixel of 151 and 133, respectively (Sun et al., 8 May 2026). CARMENES transmission spectroscopy added one full transit with VIS coverage from 0.52–0.96 $107.9^{+2.0}_{-1.7}{}^\circ$7m at $107.9^{+2.0}_{-1.7}{}^\circ$8 and NIR coverage from 0.96–1.71 $107.9^{+2.0}_{-1.7}{}^\circ$9m at $T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$0 (Gan et al., 7 Dec 2025). Dayside emission was then observed with CARMENES and PEPSI on pre-eclipse and post-eclipse nights, enabling high-resolution thermal-emission cross-correlation and retrievals (Guo et al., 25 Dec 2025).

Across these studies, the analysis architecture is characteristic of HRCCS. Spectra are shifted into the barycentric or stellar rest frame, low-S/N and telluric regions are masked, and quasi-stationary stellar and telluric structure is removed with order-by-order normalization plus PCA- or SYSREM-like detrending. MAROON-X detrending used masking of the telluric O$T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$1 B and A bands, continuum alignment, polynomial division by the median out-of-transit spectrum, and PCA along the time axis with removal of the first 5 components. HARPS-N and CARMENES transmission analyses also required explicit RM+CLV correction because the host star rotates near $T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$2, making the Doppler shadow a dominant contaminant if left untreated (Sun et al., 8 May 2026, Bello-Arufe et al., 2021, Gan et al., 7 Dec 2025).

A distinctive methodological point in the MAROON-X study is that the same detrending was applied to forward models before cross-correlation and retrieval, specifically to avoid abundance biases induced by filtering. The cross-correlation framework is based on order-summed CCFs computed over wide velocity ranges and then co-added on a $T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$3–$T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$4 grid. In the MAROON-X analysis, the CCF is

$T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$5

and the likelihood for time-series retrieval is written as

$T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$6

This formalism enabled direct retrieval on the full HRCCS time series rather than post hoc interpretation of detection maps alone (Sun et al., 8 May 2026).

3. Spectroscopic inventory of atoms, ions, and hydrides

The transmission spectrum of HAT-P-70b exhibits a particularly extensive metal inventory. The HARPS-N study detected Ca II, Cr I, Cr II, Fe I, Fe II, H I, Mg I, Na I, and V I, with tentative evidence for Ca I and Ti II (Bello-Arufe et al., 2021). MAROON-X expanded this inventory to 14 secure detections: Fe I, Fe II, Ti I, Ca I, Ca II, Cr I, Na I, V I, Mn I, Ni I, Mg I, Ba II, O I, and Sr I, with tentative evidence for H I, Co I, and K I. The strongest MAROON-X signals were Fe I at SNR $T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$7, Fe II at 19.0, the combined “All” template at 19.0, Na I at 11.5, Ca I at 10.9, and Ca II at 10.7 (Sun et al., 8 May 2026).

The red-optical and near-infrared transmission follow-up with CARMENES confirmed H$T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$8, Na I, and Ca II, reported a new tentative K I detection, and found no significant atmospheric molecular signal in the NIR. In cross-correlation, that study robustly recovered Fe I and Ca II, both blueshifted by approximately $T_{\rm eff} = 8450^{+540}_{-690}\,\mathrm{K}$9, while K I remained tentative. It also set an upper limit on the He triplet absorption at $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$0 Å (Gan et al., 7 Dec 2025).

Dayside emission spectroscopy added a distinct but overlapping inventory. The CARMENES+PEPSI emission analysis detected Al I, AlH, Ca II, Cr I, Fe I, Fe II, Mg I, Mn I, and Ti I, with tentative signals of C I, Ca I, Na I, NaH, and Ni I. This work reported the first detection of Al I and AlH in an exoplanetary atmosphere, emphasizing that in high-temperature equilibrium chemistry on the dayside, Al I and AlH are predicted to dominate over AlO (Guo et al., 25 Dec 2025).

Taken together, these studies show that the observable atmosphere is dominated by atomic and ionic opacity rather than by molecules. This is consistent with the UHJ regime, in which strong irradiation, thermal dissociation, and thermal ionization reshape the chemical inventory in the line-forming region. A plausible implication is that HAT-P-70b occupies a thermochemical regime where transmission and emission spectra are complementary rather than redundant: transmission accentuates ionization-sensitive upper terminator layers, whereas dayside emission retains sensitivity to deeper neutral-metal-bearing regions (Sun et al., 8 May 2026, Guo et al., 25 Dec 2025).

4. Atmospheric dynamics, line formation altitude, and escape diagnostics

A persistent result across transmission studies is a net blueshift of a few $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$1, interpreted as day-to-night winds. In MAROON-X data, example offsets relative to the stellar rest frame are $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$2 for Fe I, $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$3 for Fe II, $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$4 for Na I, $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$5 for V I, $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$6 for Ca II, and $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$7 for the combined species template. The CARMENES transmission analysis obtained $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$8 for Fe I and $M_\star = 1.890^{+0.010}_{-0.013}\,M_\odot$9 for Ca II from Gaussian fits to the best-$R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$0 CCFs (Sun et al., 8 May 2026, Gan et al., 7 Dec 2025).

The HARPS-N study already showed that these shifts are species dependent. Fe I was measured at $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$1, Fe II at $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$2, Na I at $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$3, Mg I at $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$4, and V I at $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$5. The same study noted broader profiles for several species than can be explained by instrumental, thermal, rotational, and exposure-smearing terms alone, with Ca II especially broad at FWHM $R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$6. This suggests vertically stratified wind fields and/or line formation over extended pressure ranges (Bello-Arufe et al., 2021).

The line-by-line transmission spectroscopy of Ca II and H I implies that the upper atmosphere is highly extended. HARPS-N resolved the Ca II H and K lines as well as H$R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$7, H$R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$8, and H$R_\star = 1.858^{+0.119}_{-0.091}\,R_\odot$9. Ca II K showed a depth of $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$0, $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$1, and $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$2; Ca II H gave $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$3. H$[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$4 had depth $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$5, FWHM $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$6, and $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$7. These cores form well above the continuum, indicating an extended envelope and possibly ongoing escape if the true planetary mass is sufficiently low (Bello-Arufe et al., 2021).

Later CARMENES transmission spectroscopy found H$[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$8 depth $[\mathrm{Fe/H}] = -0.059^{+0.075}_{-0.088}$9, equivalent width $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$0 mÅ, and $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$1, consistent with the earlier detection. The same study did not detect He I $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$2 Å, obtaining a $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$3 upper limit of depth $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$4 and $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$5 mÅ. Using XMM-Newton-based XUV estimates, it inferred an energy-limited mass-loss rate $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$6, equivalent to approximately $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$7, while emphasizing that the He non-detection remains consistent with the expected signal strength (Gan et al., 7 Dec 2025).

5. Retrieval frameworks, abundance structure, and thermal inversion

The most detailed abundance analysis to date is the MAROON-X SCARLET retrieval, implemented with emcee and carried out in two modes: a free, vertically uniform VMR framework and a chemical-equilibrium framework using FastChem with thermal ionization. Relative abundances are expressed as

$$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$8

with Ti and V defined as atomic+molecular sums where relevant. The central result is that well-mixed neutral-only retrievals can strongly bias the inferred elemental abundances of highly ionizable species. In the free retrieval, $$v \sin i = 99.85^{+0.64}_{-0.61}\,\mathrm{km\,s^{-1}}$$9 dex and $P = 2.744320 \pm 0.000001$0 dex, apparently implying severe depletions; in the equilibrium+ionization retrieval, these become $P = 2.744320 \pm 0.000001$1 dex and $P = 2.744320 \pm 0.000001$2 dex, respectively. At $P = 2.744320 \pm 0.000001$3 K and $P = 2.744320 \pm 0.000001$4 mbar, the quoted ionization fractions are approximately 0.67 for Fe and 0.98 for Ti, directly explaining why Ti and Ca are especially sensitive to ionization-aware modeling (Sun et al., 8 May 2026).

The same MAROON-X retrieval places Mg, Mn, Cr, Co, and Ca near solar relative to Fe under the equilibrium prescription, while Ti remains slightly subsolar. It also finds $P = 2.744320 \pm 0.000001$5 in the chemical-equilibrium framework and $P = 2.744320 \pm 0.000001$6–$P = 2.744320 \pm 0.000001$7. The retrieved mass, $P = 2.744320 \pm 0.000001$8, is consistent with the earlier spectroscopic estimate (Sun et al., 8 May 2026).

The most conspicuous anomaly is nickel. MAROON-X finds $P = 2.744320 \pm 0.000001$9 dex in the free retrieval and $P = 2.74432452^{+0.00000079}_{-0.00000068}$0 dex in the equilibrium retrieval, corresponding to Ni/Fe enhanced by about a factor of 20 relative to solar in the latter case. The paper reports that this enrichment is robust to alternative line lists, is not removed by masking the strongest Ni features, and agrees qualitatively with an independent dayside-emission result of $P = 2.74432452^{+0.00000079}_{-0.00000068}$1 from CARMENES+PEPSI (Sun et al., 8 May 2026, Guo et al., 25 Dec 2025).

Dayside retrievals provide the complementary thermal structure. The CARMENES+PEPSI emission study, using petitRADTRANS with both chemical-free and chemical-equilibrium retrievals, found a strong thermal inversion. In the equilibrium retrieval, temperature rises from roughly $P = 2.74432452^{+0.00000079}_{-0.00000068}$2 K at $P = 2.74432452^{+0.00000079}_{-0.00000068}$3 bar to $P = 2.74432452^{+0.00000079}_{-0.00000068}$4 K at $P = 2.74432452^{+0.00000079}_{-0.00000068}$5 bar, with $P = 2.74432452^{+0.00000079}_{-0.00000068}$6 K across the inversion region. Iron-based metallicity remains near solar: $P = 2.74432452^{+0.00000079}_{-0.00000068}$7 in the free retrieval and $P = 2.74432452^{+0.00000079}_{-0.00000068}$8 in equilibrium. High-$P = 2.74432452^{+0.00000079}_{-0.00000068}$9 elements such as Ca, Ti, and V appear somewhat depleted, while Al remains solar-like despite its high condensation temperature (Guo et al., 25 Dec 2025).

6. Comparative interpretation and open problems

HAT-P-70b is now interpreted as a chemically rich, dynamically active UHJ in the titanium transition regime. The transmission spectrum indicates that most metal ratios relative to Fe are near solar once thermal ionization is modeled, while Ti remains mildly subsolar. The MAROON-X study treats this as evidence against complete Ti cold trapping but consistent with a partial nightside cold trap: Ti I is detected, prior work reported tentative Ti II, and recent emission spectroscopy also detects Ti I, yet $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$00 and especially $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$01 remain subsolar. This suggests partial sequestration of Ti on the nightside followed by replenishment through advection and mixing (Sun et al., 8 May 2026, Guo et al., 25 Dec 2025).

Nickel remains the major unresolved abundance problem. Several interpretations are discussed in the literature: accretion of Ni-rich planetesimals, opacity aliasing by unmodeled species, differential photoionization of Fe relative to Ni, or a more generic issue in UHJ Ni chemistry and/or line lists. The MAROON-X paper explicitly notes that current 1D equilibrium models may be missing photoionization, 3D transport, and non-equilibrium chemistry, while the dayside-emission study likewise frames the Ni signal as possibly connected to Ni-rich differentiated material but not uniquely explained by that scenario (Sun et al., 8 May 2026, Guo et al., 25 Dec 2025).

Upper-atmosphere disequilibrium is a second open issue. In the dayside chemical-free retrieval, the inferred ionic VMRs for Fe II and especially Ca II are so large that the authors regard them as inconsistent with realistic hydrostatic LTE expectations. They therefore interpret the ionic line strengths as signatures of hydrodynamic escape and NLTE effects, with hydrostatic retrieval compensating for missing expansion physics by inflating ion abundances (Guo et al., 25 Dec 2025). This interpretation aligns qualitatively with the extended Ca II and Balmer absorption seen in transmission and with the CARMENES mass-loss estimate, although the He I non-detection shows that escape diagnostics remain tracer dependent (Bello-Arufe et al., 2021, Gan et al., 7 Dec 2025).

Within the broader UHJ population, HAT-P-70b supports two empirical patterns noted in the 2025 CARMENES transmission study. First, H$1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$02 absorption is more common for gas giants orbiting stars younger than 1 Gyr, with

$1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$03

Second, among UHJs with $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$04 K where both species have been searched for, Ca II detections are accompanied by Fe I detections, with conditional probability approximately $1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$05 in the current sample. HAT-P-70b exhibits both behaviors: strong H$1.87^{+0.15}_{-0.10}\,R_{\rm Jup}$06, robust Ca II, and robust Fe I (Gan et al., 7 Dec 2025).

As an overview, HAT-P-70b is best regarded as a nearly polar A-star ultra-hot Jupiter whose atmosphere is dominated by neutral and ionized metals, shaped by thermal ionization and day-to-night circulation, marked by a strong dayside temperature inversion, and complicated by partial refractory condensation and upper-atmosphere disequilibrium. The most secure compositional conclusion is that ionization-aware retrievals move many apparently depleted species back toward solar relative abundances. The most persistent anomalies are nickel enhancement, residual titanium depletion, and unusually strong ionic signatures, all of which point to physics beyond a simple 1D hydrostatic equilibrium description (Sun et al., 8 May 2026).