[Paper Review] JWST-TST High Contrast: Achieving direct spectroscopy of faint substellar companions next to bright stars with the NIRSpec IFU

Jean-Baptiste Ruffio, Marshall D. Perrin+34 more|arXiv (Cornell University)|Oct 15, 2023

Stellar, planetary, and galactic studiesPhysics and Astronomy3 citations

TL;DR

This paper presents a novel data reduction framework enabling direct spectroscopy of faint substellar companions using the James Webb Space Telescope's NIRSpec integral field unit (IFU), achieving a sensitivity of 2×10⁻⁶ flux ratio at 1" separation. By forward modeling starlight and companion signals directly in detector images and combining reference-star PSF subtraction with non-uniform spectral extraction, the team obtained a moderate-resolution (R ~ 2,700) spectrum of the T-dwarf HD 19467 B with S/N ~ 10 per resolution element at 2.9–5.2 µm.

ABSTRACT

The JWST NIRSpec integral field unit (IFU) presents a unique opportunity to observe directly imaged exoplanets from 3-5 um at moderate spectral resolution (R~2,700) and thereby better constrain the composition, disequilibrium chemistry, and cloud properties of their atmospheres. In this work, we present the first NIRSpec IFU high-contrast observations of a substellar companion that requires starlight suppression techniques. We develop specific data reduction strategies to study faint companions around bright stars, and assess the performance of NIRSpec at high contrast. First, we demonstrate an approach to forward model the companion signal and the starlight directly in the detector images, which mitigates the effects of NIRSpec's spatial undersampling. We demonstrate a sensitivity to planets that are 3e-6 fainter than their stars at 1'', or 3e-5 at 0.3''. Then, we implement a reference star point spread function (PSF) subtraction and a spectral extraction that does not require spatially and spectrally regularly sampled spectral cubes. This allows us to extract a moderate resolution (R~2,700) spectrum of the faint T-dwarf companion HD 19467 B from 2.9-5.2 um with signal-to-noise ratio (S/N)~10 per resolution element. Across this wavelength range, HD~19467~B has a flux ratio varying between 1e-5-1e-4 and a separation relative to its star of 1.6''. A companion paper by Hoch et al. more deeply analyzes the atmospheric properties of this companion based on the extracted spectrum. Using the methods developed here, NIRSpec's sensitivity may enable direct detection and spectral characterization of relatively old (~1 Gyr), cool (~250 K), and closely separated (~3-5 au) exoplanets that are less massive than Jupiter.

Motivation & Objective

• To enable direct spectroscopy of faint substellar companions around bright stars using the JWST NIRSpec IFU, overcoming challenges from spatial undersampling and high dynamic range.
• To develop data reduction strategies that minimize interpolation errors in highly undersampled IFU data, improving sensitivity and spectral fidelity.
• To demonstrate the feasibility of extracting moderate-resolution spectra from high-contrast imaging data using reference-star PSF subtraction and non-uniform spectral extraction.
• To establish a forward modeling approach that directly models starlight and companion signals in detector images, enhancing photon noise-limited detection sensitivity.
• To enable future detection and characterization of old, cool, close-in exoplanets (e.g., 1 Gyr, ~250 K, 3–5 au) with NIRSpec IFU.

Proposed method

• Forward modeling of both starlight and companion signal directly in detector images to mitigate effects of NIRSpec’s spatial undersampling and achieve photon noise-limited sensitivity.
• Implementation of reference-star PSF subtraction using a reference star with similar brightness and angular separation to suppress stellar halo and residual speckles.
• Development of a spectral extraction method that does not require spatially and spectrally regular sampling, enabling robust extraction from irregularly sampled IFU cubes.
• Use of the G395H grating and F290LP filter to achieve moderate spectral resolution (R ~ 2,700) across 2.9–5.3 µm.
• Application of statistical correction for small sample size in sensitivity curves, improving detection limit estimation.
• Integration of forward modeling with reference differential imaging (RDI) to optimally combine high-resolution spectral features with continuum information.

Experimental results

Research questions

• RQ1 Can the NIRSpec IFU achieve photon noise-limited detection sensitivity for substellar companions despite spatial undersampling?
• RQ2 Can a forward modeling approach in detector space outperform interpolation-based methods in high-contrast IFU data reduction?
• RQ3 What is the achievable spectral signal-to-noise ratio for a faint T-dwarf companion using NIRSpec IFU with reference-star PSF subtraction?
• RQ4 Can moderate-resolution spectroscopy (R ~ 2,700) be successfully extracted from high-contrast IFU data for a companion with flux ratio ~10⁻⁵–10⁻⁴?
• RQ5 To what extent can the combination of forward modeling and RDI improve detection and characterization sensitivity for exoplanets?

Key findings

• The team achieved a detection sensitivity of 2×10⁻⁶ flux ratio at 1" separation, corresponding to a signal-to-noise ratio of ~249 for HD 19467 B in 35 minutes of integration.
• A moderate-resolution spectrum of HD 19467 B was extracted between 2.9–5.2 µm with a continuum signal-to-noise ratio of ~10 per resolution element.
• The flux ratio of HD 19467 B varies between 10⁻⁵ and 10⁻⁴ at 1.6" separation, consistent with a T-dwarf companion at ~250 K and ~1 Gyr age.
• The forward modeling approach in detector images enabled photon noise-limited sensitivity, reducing errors from spatial interpolation.
• Interpolation errors in the RDI implementation were identified as the primary limitation on spectral signal-to-noise, suggesting that minimizing interpolation is critical for high-contrast IFU science.
• The proposed method combining forward modeling and RDI shows promise for improving detection and characterization sensitivity for future high-contrast exoplanet studies with NIRSpec IFU.