Direct-method metallicity gradients derived from spectral stacking with SDSS-IV MaNGA

Chemical abundances are key tracers of the cycle of baryons that drives the evolution of galaxies. Most measurements of interstellar medium abundances and metallicity gradients in galaxies are, however, based on model-dependent strong-line methods. Direct chemical abundances can be obtained via the detection of weak auroral lines, but such lines are too faint to be detected by large spectroscopic surveys of the local Universe. In this work we overcome this limitation and obtain metallicity gradients from direct-method abundances by stacking spectra from the MaNGA integral field spectroscopy survey. In particular, we stacked 4140 star-forming galaxies across the star formation rate-stellar mass plane and across six radial bins. We calculated electron temperatures for [OII], [SII], [NII], [SIII], and [OIII] across the majority of the stacks. We find that T[OII] approximate to T[SII] approximate to T[OII], as expected since these ions all trace the low-ionization zone of nebulae. The [OIII] temperatures become substantially higher than those of other ions at high metallicities, indicating potentially unaccounted for spectral contamination or additional physics. In light of this uncertainty, we based our abundance calculation on the temperatures of [SIII] and the low-ionization ions. We recover a mass-metallicity relation similar to that obtained with different empirical calibrations. We do not, however, find evidence of a secondary dependence on the star formation rate using direct metallicities. Finally, we derive metallicity gradients that become steeper with stellar mass for log(M-star/M-circle dot) < 10.5. At higher masses, the gradients flatten again, confirming with auroral line determinations the trends previously defined with strong-line calibrators.