Context :

Aims : Previous high-precision studies of abundances of elements in solar twin stars are extended to a wider metallicity range to see how the trends of element ratios with stellar age depend on [ Fe/H ] .

Methods : HARPS spectra with signal-to-noise ratios S / N \lower 2.15 pt \hbox { $ \buildrel > \over { \sim } $ } 600 at \lambda \sim 6000 Å were analysed with MARCS model atmospheres to obtain 1D LTE abundances of C , O , Na , Mg , Al , Si , Ca , Ti , Cr , Fe , Ni , Sr , and Y for 72 nearby solar-type stars with metallicities in the range -0.3 \lower 2.15 pt \hbox { $ \buildrel < \over { \sim } $ } \mbox { [ Fe / H ] } \lower 2.15 % pt \hbox { $ \buildrel < \over { \sim } $ } +0.3 and ASTEC stellar models were used to determine stellar ages from effective temperatures , luminosities obtained via \it Gaia DR2 parallaxes , and heavy element abundances .

Results : The age-metallicity distribution appears to consist of two distinct populations : a sequence of old stars with a steep rise of [ Fe/H ] to \sim +0.3 dex at an age of \sim 7 Gyr and a younger sequence with [ Fe/H ] increasing from about -0.3 dex to \sim +0.2 dex over the last six Gyr .
Furthermore , the trends of several abundance ratios , [ O/Fe ] , [ Na/Fe ] , [ Ca/Fe ] , and [ Ni/Fe ] , as a function of stellar age split into two corresponding sequences .
The [ Y/Mg ] -age relation , on the other hand , shows no offset between the two age sequences and has no significant dependence on [ Fe/H ] , but the components of a visual binary star , \zeta Reticuli , have a large and puzzling deviation .

Conclusions : The split of the age-metallicity distribution into two sequences may be interpreted as evidence of two episodes of accretion of gas onto the Galactic disk with a quenching of star formation in between .
Some of the [ X/Fe ] -age relations support this scenario but other relations are not so easy to explain , which calls for a deeper study of systematic errors in the derived abundances as a function of [ Fe/H ] , in particular 3D non-LTE effects .