Digest
Pang et al. build a slitless-spectrum QSO selector on 3D-HST/G141 by first flagging >6,000 sources with apparently broad lines and then using forward modeling to test whether the “broad” features require a point-like QSO component rather than size-broadened ELG lines. Calibrated on known QSOs, ELGs, and simulations, the criterion recovers ~90% of point-like QSOs with Hα/Hβ while keeping ELG contamination at ~5%. Applied to clean emission-line sources, it yields 19 QSO candidates at z=0.12–1.56 (12 X-ray detections) with broader UV colors and redder optical slopes than SDSS QSOs and MBH≈10^6.9–10^8.3 M⊙, consistent with normal QSOs rather than little red dots. The method boosts QSO completeness at z≈0.8–1.6 in 3D-HST and is directly portable to Euclid and CSST slitless surveys.
Key figures to inspect
- Figure 1 — Walk through the two-step pipeline; focus on how the forward-model “QSO component reliability” gate is defined and where size-broadened ELG lines fail the test.
- Figure 2 — Inspect the continuum-fit and thresholding on 1D spectra; compare the distorted versus clean examples to see how peaks are flagged and how apparent FWHM≥1200 km/s are distinguished from noise.
- Figure 3 — Examine heavy-contamination cases in 2D spectra; compare the data, model, and contamination model to see how spurious broadening from overlaps is recognized and removed.
- Figure 4 — Check the F140W–redshift distribution for the one-line vs two-line subsamples; note that most lie at 0.6–2.4 where Hα or Hβ+[O III] enters G141, setting the method’s most effective redshift window.