We investigate the enrichment in elements produced by the slow neutron-capture process ($s$-process) in the globular clusters M4 (NGC 6121) and M22 (NGC 6656). Stars in M4 have homogeneous abundances of Fe and neutron-capture elements, but the entire cluster is enhanced in $s$-process elements (Sr, Y, Ba, Pb) relative to other clusters with a similar metallicity. In M22, two stellar groups exhibit different abundances of Fe and $s$-process elements. By subtracting the mean abundances of $s$-poor from $s$-rich stars, we derive $s$-process residuals or empirical $s$-process distributions for M4 and M22. We find that the $s$-process distribution in M22 is more weighted toward the heavy $s$-peak (Ba, La, Ce) and Pb than M4, which has been enriched mostly with light $s$-peak elements (Sr, Y, Zr). We construct simple chemical evolution models using yields from massive star models that include rotation, which dramatically increases $s$-process production at low metallicity. We show that our massive star models with rotation rates of up to 50% of the critical (break-up) velocity and changes to the preferred $^{17}$O($alpha$,$gamma$)$^{21}$Ne rate produce insufficient heavy $s$-elements and Pb to match the empirical distributions. For models that incorporate AGB yields, we find that intermediate-mass yields (with a $^{22}$Ne neutron source) alone do not reproduce the light-to-heavy $s$-element ratios for M4 and M22, and that a small contribution from models with a $^{13}$C pocket is required. With our assumption that $^{13}$C pockets form for initial masses below a transition range between 3.0 and 3.5 M$_odot$, we match the light-to-heavy s-element ratio in the s-process residual of M22 and predict a minimum enrichment timescale of between 240 and 360 Myr. Our predicted value is consistent with the 300 Myr upper limit age difference between the two groups derived from isochrone fitting.
Download