Extended Data Fig. 7: Average momentum space distributions.

a, b, The mean momentum-space distributions n_↑(p_↓) of a single spin component and averaged over 1,000 images for two different binding energies and 6 + 6 atoms are shown. The dashed circle indicates the Fermi momentum. The distributions are to a good approximation radially symmetric. With increasing binding energy, the average momentum increases and we find more particles outside the Fermi momentum. This agrees with the picture that increasing the attraction enables particles to overcome the single-particle gap and form first Cooper pairs that finally turn into tightly bound dimers. c, From the average momentum-space distributions of both spin components it is straightforward to calculate the total mean kinetic energy of the system per spin component. For 6 + 6 non-interaction particles, we find a value very close to the expected ground-state (GS) kinetic energy per spin component of \({E}_{{\rm{kin}}}^{{\rm{gs}}}=7\hbar {\omega }_{r}\). The kinetic energy increases monotonously as the attraction strength increases. The error bars represent the standard error of the mean. d, The unnormalized correlator \({{\mathscr{C}}}_{{\bar{p}}_{\downarrow }}^{(2{)}^{\ast }}\) is shown for E_B/ħω_r = 1.97. It is defined as in equations (1), (2) but without the ⟨n⟩⟨n⟩ term. It shows that for strong enough binding energies the paired fraction becomes large enough that pair correlations are visible even without subtracting the single-particle density contributions.