Ultracold Dipolar Molecules Composed of Strongly Magnetic Atoms

In a combined experimental and theoretical effort, we demonstrate a novel type of dipolar system made of ultracold bosonic dipolar molecules with large magnetic dipole moments. Our dipolar molecules are formed in weakly bound Feshbach molecular states from a sample of strongly magnetic bosonic erbium atoms. We show that the ultracold magnetic molecules can carry very large dipole moments and we demonstrate how to create and characterize them, and how to change their orientation. Finally, we confirm that the relaxation rates of molecules in a quasi-two-dimensional geometry can be reduced by using the anisotropy of the dipole-dipole interaction and that this reduction follows a universal dipolar behavior.

Figure 1

${\mathrm{Er}}_2$ weakly bound molecules. (a) Atom-loss spectrum [4] from 0 to 3 G and (b) near-threshold binding energy of the corresponding molecular states. The solid lines are fits to the experimental data and extrapolated to larger $E_b$ up to $h\times 500\mathrm{kHz}$. The error bars are smaller than the symbols.

Figure 2

Adiabatic magnetic moment as a function of magnetic-field strength evaluated at the entrance channel energy. Each curve corresponds to the adiabatic magnetic moment of one adiabatic potential $U_n(R;B)$. The magnetic moments in the asymptotic limit of large $B$ are given. The dashed vertical lines correspond to the field strength where we have observed Feshbach resonances. The red-filled circles represent the experimentally measured magnetic moments at these resonance locations.

Figure 3

Typical time evolution of the number of molecules for $\theta =90°$ (squares) and $\theta =0°$ (circles) in a 3D (a) and in a Q2D trap (b). The data refer to molecules in the state ${\mu }_1$ for the 3D case (a) and molecules in the state ${\mu }_2$ in Q2D (b). The insets in (b) show an illustration of molecules in pancake-shaped traps with out-of-plane (right) and in-plane (left) orientations. The solid lines are two-body decay fits to the data. The error bars for (a) and (b) are smaller than the data points and are not shown. The data points in (a) are obtained by averaging five independent measurements and in (b) about 50 measurements have been averaged.

Figure 4

Universal loss rate ratio ${\beta }_{\perp }/{\beta }_{\parallel }$ as a function of $a_d/\tilde{a}$ for ${\mathrm{Er}}_2$ (circles) and KRb (squares) for a fixed value $a_{\mathrm{dB}}/a_{\mathrm{ho}}=4.85$ corresponding to $T=400(40)\mathrm{nK}$ and ${\nu }_z=31.2(1)\mathrm{kHz}$ [16]. The gray shaded area is due to the uncertainty of $T$. Here, $\tilde{a}=a_{\mathrm{ho}}$ for bosonic molecules (filled symbols) and $\tilde{a}=a_{\mathrm{vdW}}$ for fermionic molecules (open symbols). The calculated loss rate ratios of ${\mathrm{Er}}_2$ are compared with the experimental data for states ${\mu }_1$, ${\mu }_2$, and ${\mu }_4$ (triangles).