The atom chip used in our experiment was designed and built by Philipp Treutlein and Pascal B¨ohi. Two features, employed here for the first time, make it one of the technically most advanced atom chips in existence: first, two layers of gold, separated by a thin insulating layer, allow for crossing wires and thus more flexibility in trap design. Second, integrated coplanar waveguides (CPWs) allow us to inject microwave currents into the chip and use the CPW near-field for state-selective manipulation of the atoms.
A photograph of the chip is shown in figure 3.1 and a drawing of the wire layout in figure 3.2. An AlN base chip provides mechanical stability and
Experimental setup base chip science chip gold m irror microwave connector DC a da p te r on t he ba ck CPW bond wires 50 mm
Figure 3.1: Microwave atom chip. Microwave atom chip assembly, in- cluding the base chip, the science chip, and electrical DC and microwave connectors, before it is glued to the vacuum glass cell.
easy connection to DC current sources through the socket adapters, which are soldered onto the chip from the back. Microwave sources are connected from the front. The electroplated gold wires have a height of 12µm and can carry DC currents of up to 10 A.
A spacer chip is glued onto the base chip, and the so called science chip
is glued onto this spacer chip. Both have the same dimensions and are cut from a 525µm thick high-resistivity Silicon wafer. The science chip carries two layers of metalization, separated by a layer of polyimide which is 6µm thick. The polyimide provides not only electrical insulation between the two gold layers but also planarization to reduce the bumpiness of the upper layer. The lower gold layer has a thickness of 5µm and is fabricated in the same way as the wires on the base chip. The wires on this layer carry static currents for magnetic Ioffe-Pritchard and dimple traps. The upper layer is fabricated with a lift-off technique and has a thickness of only 1µm. Some wires on this layer form CPWs but can also carry DC currents of up to about 100 mA. In the experiment reported here, we use a 5-wire structure. The central three wires form the CPW indicated in figure 3.2 while the outer two wires can be used for additional tuning of the position of the static magnetic trap. The upper layer also features a large gold mirror which is used for the mirror MOT in the first stages of the experimental cycle. Base and science chip are electrically connected through gold bond wires.
3.1 Microwave atom chip CPW CPW CPW ID ID ID ILI ILI IZ IZ IZ IZ + ILI + ILI x z y 50 mm 1 mm
Figure 3.2: Wire layout. The wire layout on the base chip (green) and the lower (blue) and upper layer (red) of the science chip. The right panel shows a zoom into the experiment region, omitting the gold mirror. There are several waveguide structures on the chip, but for the experiments reported in this thesis, only the one indicated is used. The black arrows indicate the main DC currents used in our experimental sequence, as described in section 3.8.
3.1.1
Characterization of the microwave near-field
Using the Biot-Savart law, it is straightforward to calculate magnetic fields produced by static currents and thus simulate the expected magnetic trap po- sitions and frequencies with high accuracy (within a few percent). The simu- lation of the microwave near-field proves to be much more difficult. Since the transverse CPW dimensions and the distance of the atoms from the wires are much smaller than the microwave wavelength (λ= 4.4 cm forfmw = 6.8 GHzin vacuum), we can neglect retardation effects and calculate the microwave field around the CPW from the microwave current distribution in the wires in a similar way to the static fields. The difficulty lies in determining the exact current distribution, since microwave currents can be induced in the adjacent wires or the wires on the lower layer of the science chip. This and the curva- ture of the CPW can furthermore lead to an asymmetric current distribution in the CPW itself. Also, the skin effect leads to an inhomogeneous current distribution in each wire.
We have devised a method to measure the microwave field distribution around the waveguide, using the cold atoms themselves as probes [64, 45].
Experimental setup
Briefly, we release a cloud of thermal atoms in state|0⟩close to the waveguide and let it expand until it fills an extended region around the guide. We then apply a short microwave pulse in the CPW, resonant with an atomic tran- sition. We thus coherently transfer some population into the F = 2 states. The transfer rate at any given point in space depends on the microwave field strength at that point which therefore can be deduced by state-selective de- tection of the atoms. The data quality can be improved by scanning the microwave amplitude to record Rabi oscillations in space. By repeating such measurements with a static magnetic quantization field pointing along each coordinate axis, and with the microwave frequency adjusted to drive the π,
σ+, or σ− transition (for a total of 9 data sets), we can reconstruct the
complete microwave field distribution in the region filled by the atoms. We compare the measured field to the field calculated from the assumed current distribution, and adjust the latter to maximize agreement. A good match can be achieved by assuming a small asymmetry in the microwave currents on the CPW and small induced currents in the two adjacent wires as well as in the lower science chip layer. We have also simulated the mi- crowave propagation on the CPW, including the lower science chip layer, using the software Sonnet and find good agreement between this simulation and the experimentally found current distribution. As an example, figure 3.3 shows the measured and simulated distributions of the x-component of the microwave magnetic field.