Transport of atom packets in a train of Ioffe-Pritchard traps

We demonstrate transport and evaporative cooling of several atomic clouds in a chain of magnetic Ioffe-Pritchard traps moving at a low speed $(<1\mathrm{m}∕\mathrm{s})$. The trapping scheme relies on the use of a magnetic guide for transverse confinement and of magnets fixed on a conveyor belt for longitudinal trapping. This experiment introduces a different approach for parallelizing the production of Bose-Einstein condensates as well as for the realization of a continuous atom laser.

Figure 1

(Color online) Creation of a longitudinal confinement with permanent magnets. (a) Magnets with alternating magnetization directions (arrows) located on a line parallel to $z$ create a modulated $z$ component of the magnetic field (solid line). A constant offset field $B_{\mathrm{sol}}$ (dashed line) can be added in order to keep a constant sign for $B_z$. (b) The potential $U$ experienced by the atoms for $B_{\mathrm{sol}}=0$ is a chain of quadrupole traps (top curve). Due to the transverse component $B_x$ of the field, the traps are shifted from the guide axis, as can be seen on the contour plot of $\mid \mathit{B}\mid$ in the $(x,z)$ plane. The isocontours are plotted every $20\mathrm{G}$, starting at $0\mathrm{G}$. (c) With a large enough $B_{\mathrm{sol}}$, one gets a chain of on-axis Ioffe-Pritchard traps.

Figure 2

(Color online) Experimental setup.

Figure 3

(Color online) Absorption of a probe beam located at $z_2=1.7\mathrm{m}$ from the guide entrance as a function of time $t$ elapsed after the launch of a single atom packet, at the velocity $v_{\mathrm{i}}=80\mathrm{cm}∕\mathrm{s}$ (with a rms velocity dispersion of $20\mathrm{cm}∕\mathrm{s}$). (a) The IP traps have a depth of $50\mathrm{G}$ and a minimum field of $1\mathrm{G}$, and one varies $v_{\mathrm{conv}}$. A sharp peak of trapped atoms arrives at a time $t_{\mathrm{p}}$. When the velocity mismatch between the conveyor and the injected atoms is too large, extra peaks corresponding to untrapped atoms appear (⋆). (b) Arrival time $t_{\mathrm{p}}$ of the trapped atoms (squares) as a function of $v_{\mathrm{conv}}$; $t_{\mathrm{p}}$ is essentially equal to $z_2∕v_{\mathrm{conv}}$ (dashed line). (c) The conveyor speed is fixed at $50\mathrm{cm}∕\mathrm{s}$ and one varies the IP trap depth. For too shallow traps, one gets a peak of untrapped atoms (⋆); if the depth is too high, some atoms are reflected, which decreases the peak amplitude.

Figure 4

Simulated atomic distribution in phase space $(z,v_z)$ for various times $\tau$ after the launch of the packet. The initial velocity is $v_{\mathrm{i}}=60\mathrm{cm}∕\mathrm{s}$, with a $20\mathrm{cm}∕\mathrm{s}$ dispersion, and the initial size of the packet is $5\mathrm{cm}$. Left: free propagation (no conveyor), showing the spatial spreading of the packet and the building of correlations between position and velocity. Right: the conveyor is made of IP traps with $10\mathrm{G}$ height, moving at $v_{\mathrm{conv}}=60\mathrm{cm}∕\mathrm{s}$. At $\tau =1.5\mathrm{s}$, the slow (i), trapped (ii), and fast (iii) atoms are easily identified. At $\tau =3\mathrm{s}$, the atom packets released from the conveyor have already started to spread out and overlap.

Figure 5

(Color online) Absorption signal (thick line) measured in the conveyor belt, with a probe located at $z_1=1\mathrm{m}$. The IP traps have a depth of $32\mathrm{G}$, and one has $v_{\mathrm{conv}}=v_{\mathrm{i}}=88\mathrm{cm}∕\mathrm{s}$. The thin line is the measured $B_z$ field component produced by the conveyor. (a) Proper synchronization of the launching of one packet to load a single trap (the cloud arrives in coincidence with a potential minimum). (b) The launching occurs at a time $\Lambda ∕\left(2v_{\mathrm{conv}}\right)$ before, resulting in a splitting of the cloud into two consecutive traps. (c) Multiple injection of packets, with the proper synchronization to load one every second trap.