Analyzing low signal-to-noise FUSE spectra

Confirmation of Lyman continuum escape from Haro 11^⋆

¹ Department of Physics and AstronomyUppsala University, Box 515, 751 20 Uppsala, Sweden
e-mail: elisabet.leitet@fysast.uu.se; nils.bergvall@fysast.uu.se; nikolai.piskunov@fysast.uu.se
² Stratospheric Observatory for Infrared Astronomy, NASA Ames Research Center, Mail Stop 211-3, Moffett Field, CA 94035, USA
e-mail: bgandersson@sofia.usra.edu

Received: 27 August 2010
Accepted: 26 May 2011

Abstract

Context. Galaxies are believed to be the main providers of Lyman continuum (LyC) photons during the early phases of the cosmic reionization. Little is known however, when it comes to escape fractions and the mechanisms behind the leakage. To learn more, one may look at local objects, but so far only one low-z galaxy has shown any signs of emitting LyC radiation. With data from the Far Ultraviolet Spectroscopic Explorer (FUSE), we previously found an absolute escape fraction of ionizing photons (f_esc) of 4–10% for the blue compact galaxy Haro 11. However, using a revised version of the reduction pipeline on the same data set, Grimes and collaborators were unable to confirm this and derived an upper limit of f_esc ≲ 2% .

Aims. We attempt to determine whether Haro 11 is emitting ionizing radiation to a significant level or not. We also investigate the performance of the reduction pipeline for faint targets such as Haro 11, and introduce a new approach to the background subtraction.

Methods. The final version of the reduction pipeline, CalFUSE v3.2, was applied to the same Haro 11 data set as the two previous authors used. At these faint flux levels, both FUSE and CalFUSE are pushed to their limits, and a detailed analysis was undertaken to monitor the performance of the pipeline. We show that non-simultaneous background estimates are insuffient when working with data of low signal-to-noise ratio (S/N), and a new background model was developed based on a direct fit to the detector response.

Results. We find that one has to be very careful when using CalFUSE v3.2 on low S/N data, and especially when dealing with sources where signal might originate from off-center regions. Applying the new background fit, a significant signal is detected in the LyC in both detector segments covering these wavelengths. Thus, the leakage is confirmed with a flux density of f₉₀₀ = 4.0 × 10^-15 erg s^-1 cm^-2 Å^-1 (S/N = 4.6), measured on the airglow free regions in the LyC for the night-only data. This corresponds to an absolute escape fraction of ionizing photons from Haro 11 of f_esc = 3.3 ± 0.7%. We confirm these results by investigating the two-dimensional data, the count rates, and the residual flux in C ii λ1036 Å.