The four directly imaged planets orbiting the star HR 8799 are an ideal laboratory to probe atmospheric physics and formation models. We present more than a decades worth of Keck/OSIRIS observations of these planets, which represent the most detailed
look at their atmospheres to-date by its resolution and signal to noise ratio. We present the first direct detection of HR 8799 d, the second-closest known planet to the star, at moderate spectral resolution with Keck/OSIRIS (K-band; R~4,000). Additionally, we uniformly analyze new and archival OSIRIS data (H and K band) of HR 8799 b, c, and d. First, we show detections of water (H2O) and carbon monoxide (CO) in the three planets and discuss the ambiguous case of methane (CH4) in the atmosphere of HR 8799b. Then, we report radial velocity (RV) measurements for each of the three planets. The RV measurement of HR 8799 d is consistent with predictions made assuming coplanarity and orbital stability of the HR 8799 planetary system. Finally, we perform a uniform atmospheric analysis on the OSIRIS data, published photometric points, and low resolution spectra. We do not infer any significant deviation from to the stellar value of the carbon to oxygen ratio (C/O) of the three planets, which therefore does not yet yield definitive information about the location or method of formation. However, constraining the C/O ratio for all the HR 8799 planets is a milestone for any multiplanet system, and particularly important for large, widely separated gas giants with uncertain formation processes.
$epsilon$~Eridani is a young planetary system hosting a complex multi-belt debris disk and a confirmed Jupiter-like planet orbiting at 3.48 AU from its host star. Its age and architecture are thus reminiscent of the early Solar System. The most recen
t study of Mawet et al. 2019, which combined radial velocity (RV) data and Ms-band direct imaging upper limits, started to constrain the planets orbital parameters and mass, but are still affected by large error bars and degeneracies. Here we make use of the most recent data compilation from three different techniques to further refine $epsilon$~Eridani~bs properties: RVs, absolute astrometry measurements from the Hipparcos~and Gaia~missions, and new Keck/NIRC2 Ms-band vortex coronagraph images. We combine this data in a Bayesian framework. We find a new mass, $M_b$ = $0.66_{-0.09}^{+0.12}$~M$_{Jup}$, and inclination, $i$ = $77.95_{-21.06}^{circ+28.50}$, with at least a factor 2 improvement over previous uncertainties. We also report updated constraints on the longitude of the ascending node, the argument of the periastron, and the time of periastron passage. With these updated parameters, we can better predict the position of the planet at any past and future epoch, which can greatly help define the strategy and planning of future observations and with subsequent data analysis. In particular, these results can assist the search for a direct detection with JWST and the Nancy Grace Roman Space Telescopes coronagraph instrument (CGI).
Using the Keck Planet Imager and Characterizer (KPIC), we obtained high-resolution (R$sim$35,000) $K$-band spectra of the four planets orbiting HR 8799. We clearly detected water{} and CO in the atmospheres of HR 8799 c, d, and e, and tentatively det
ected a combination of CO and water{} in b. These are the most challenging directly imaged exoplanets that have been observed at high spectral resolution to date when considering both their angular separations and flux ratios. We developed a forward modeling framework that allows us to jointly fit the spectra of the planets and the diffracted starlight simultaneously in a likelihood-based approach and obtained posterior probabilities on their effective temperatures, surface gravities, radial velocities, and spins. We measured $vsin(i)$ values of $10.1^{+2.8}_{-2.7}$~km/s for HR 8799 d and $15.0^{+2.3}_{-2.6}$~km/s for HR 8799 e, and placed an upper limit of $< 14$~km/s of HR 8799 c. Under two different assumptions of their obliquities, we found tentative evidence that rotation velocity is anti-correlated with companion mass, which could indicate that magnetic braking with a circumplanetary disk at early times is less efficient at spinning down lower mass planets.
The Keck Planet Imager and Characterizer (KPIC) is an upgrade to the Keck II adaptive optics system and instrument suite with the goal of improving direct imaging and high-resolution spectroscopic characterization capabilities for giant exoplanets. K
PIC Phase I includes a fiber injection unit (FIU) downstream of a new pyramid wavefront sensor, coupling planet light to a single mode fiber fed into NIRSPEC, Kecks high-resolution infrared spectrograph. This enables high-dispersion spectroscopy (HDS) of directly imaged exoplanets at smaller separation and higher contrast, improving our spectral characterization capabilities for these objects. Here, we report performance results from the KPIC Phase I FIU commissioning, including analysis of throughput, stability, and sensitivity of the instrument.
High-contrast medium resolution spectroscopy has been used to detect molecules such as water and carbon monoxide in the atmospheres of gas giant exoplanets. In this work, we show how it can be used to derive radial velocity (RV) measurements of direc
tly imaged exoplanets. Improving upon the traditional cross-correlation technique, we develop a new likelihood based on joint forward modelling of the planetary signal and the starlight background (i.e., speckles). After marginalizing over the starlight model, we infer the barycentric RV of HR 8799 b and c in 2010 yielding -9.2 +- 0.5 km/s and -11.6 +- 0.5 km/s respectively. These RV measurements help to constrain the 3D orientation of the orbit of the planet by resolving the degeneracy in the longitude of ascending node. Assuming coplanar orbits for HR 8799 b and c, but not including d and e, we estimate Omega = 89 (+27,-17) deg and i = 20.8 (4.5,-3.7) deg.
We present a new matched filter algorithm for direct detection of point sources in the immediate vicinity of bright stars. The stellar Point Spread Function (PSF) is first subtracted using a Karhunen-Loeve Image Processing (KLIP) algorithm with Angul
ar and Spectral Differential Imaging (ADI and SDI). The KLIP-induced distortion of the astrophysical signal is included in the matched filter template by computing a forward model of the PSF at every position in the image. To optimize the performance of the algorithm, we conduct extensive planet injection and recovery tests and tune the exoplanet spectra template and KLIP reduction aggressiveness to maximize the Signal-to-Noise Ratio (SNR) of the recovered planets. We show that only two spectral templates are necessary to recover any young Jovian exoplanets with minimal SNR loss. We also developed a complete pipeline for the automated detection of point source candidates, the calculation of Receiver Operating Characteristics (ROC), false positives based contrast curves, and completeness contours. We process in a uniform manner more than 330 datasets from the Gemini Planet Imager Exoplanet Survey (GPIES) and assess GPI typical sensitivity as a function of the star and the hypothetical companion spectral type. This work allows for the first time a comparison of different detection algorithms at a survey scale accounting for both planet completeness and false positive rate. We show that the new forward model matched filter allows the detection of $50%$ fainter objects than a conventional cross-correlation technique with a Gaussian PSF template for the same false positive rate.
