Cryogenic system

Cryogenic techniques are in the first place required to cool the helium buffer-gas.

As has been pointed out in Section 2.3.4, temperatures below 80 mK should be accessible and high cooling powers are favorable for a complete and quick buffer-gas removal. Further considerations on the design of a suitable cryostat should include helium consumption, the available operating time between refills, and a

36

CRYOGENIC SYSTEM 37 convenient assembly of all parts. All of these issues have to be made compatible with the installation of a superconducting magnet and sufficient optical access to its field minimum.

Based on mature technology, a new³He-⁴He dilution refrigerator has specifi-cally been developed for these purposes, which is described in detail in a second thesis related to this project [82]. Only a brief summary of its specifications shall therefore be given here.

3.1.1 Dilution refrigerator configuration

Main components of the initial configuration of the dilution refrigerator are illustrated to scale in Figure 3.1. The cryostat itself was built by VOSKAM with a design goal of 72 h of continuous operation between helium refills. It has a total height of approximately 240 cm and a maximum diameter of about 50 cm. In the outer vacuum chamber (OVC), a liquid nitrogen (LN) ring tank surrounds a large liquid helium (LHe) dewar, which is connected by a total of 7 tubes to a lower tank containing the superconducting magnet. Its clear bore is part of the inner vacuum chamber (IVC) with the dilution unit, a standard Leiden Cryogenics Minikelvin 126-700 with customized cut-outs on the vertical optical axis. It has a specified cooling power of 700W at 120 mK and a base temperature below 8 mK.

Optical access to the trapping region at the magnet center is available from six sides through broadband antireflection coated fused silica windows at room temperature, 77 K, and 4 K, respectively. With a diameter of 3”, the largest aperture is offered from the bottom. On the same axis, the top port has a free diameter of 20 mm. It lacks a window at the 77 K level, which is not accessible there. Due to stability restrictions, side access through the magnet is limited to 10 mm. While these features are still quite unique for a dilution refrigerator, the overall optical access appears rather limited when compared to standard atom optics experiments. It is however more than sufficient for buffer-gas loading, as it does not require lasers for trapping, but only for sample detection. The present setup is already suited to allow the future transfer of a trapped ensemble to an optical dipole trap, for example.

The whole cryostat is supported by a versatile aluminum rack (ITEM), that can directly be attached to an accordingly fitted optical table (Figure 3.2).

Any detection optics thus has a fixed position with respect to the experimental cell even when the table is floated, thereby potentially reducing environmental noise in a measurement. The vacuum cans and radiative shielding have been sectioned such that an assembly and disassembly of the cryostat is possible without taking it out of its rack. All cryogenic flanges are sealed with indium wire of appropriate thickness, while the room-temperature connections rely on Viton o-rings.

38 EXPERIMENTAL REALIZATION

LN ring tank

80l LHe dewar

1K Pot Still

Mixing Chamber

Magnet

77K shield OVC pressure gauge

OVC evacuation Dewar safety valve 1K pot pump line

1K pot fill line with needle valve

IVC

3He circulation

3He recondensation line

Magnet tank Support

Figure 3.1: Main parts of the original dilution refrigerator.

CRYOGENIC SYSTEM 39

Figure 3.2: Dilution refrigerator picture gallery. From left to right: partially opened fridge in its rack on the optical table, bottom view with disconnected magnet tank, and Leiden Cryogenics dilution unit.

3.1.2 ³He circulation

The ³He circulation for dilution refrigeration is driven by a large turbomolecu-lar pump (Pfeiffer Vacuum TMH 1600 MC) directly flanged to the top of the cryostat. It is backed by a helium-tight rotary vane pump (Pfeiffer Vacuum DUO 20). The gas flow is organized by a commercial Leiden Cryogenics gas handling system which accompanied the dilution unit. It can be bypassed with a large diameter tube between the two pumps for higher throughput to achieve an optimal cooling efficiency. Details of the complete dilution and vacuum in-frastructure are summarized in Appendix A.

3.1.3 Thermometry

Temperatures are measured throughout using either platinum resistors (Pt-100) or 1 kΩ ruthenium oxide sensors (LakeShore RX-102A). While Pt-100s can be individually calibrated down to the boiling point of liquid helium at 4.2 K, the ruthenium oxide (ROX) sensors provide superior sensitivity from 40 K to below 50 mK, so that the total range of interest is covered. Both types are still useful in strong magnetic fields.

The sensors are read out by an AC resistance bridge (Picowatt AVS-47) with a four-terminal measurement at typical excitation voltages of 300V or less.

The resistance bridge is controlled via GPIB by a Labview program developed in this thesis. It converts the resistance readings into temperature using the

40 EXPERIMENTAL REALIZATION following fits to calibration data:

T(R) =

c₀+c₁R+c₂R²+c₃R³+c₄R⁴1/n for Pt-100 [83] and T(R) =T₀·(lnR/R₀)⁻ⁿ for ROX [71],

T is the temperature in Kelvin and R the resistance is in Ohm. To minimize the relative rather than the absolute deviation over the whole temperature range, a least square fit was applied to logarithmic values.

A prerequisite for reliable temperature readings is an appropriate thermal anchoring of the sensor to the object of interest. Thermal equilibrium between the two is largely supported by also contacting the electrical connection, part of which is therefore usually epoxied to a copper lug also housing the sensor. It is then attached to the object with a brass screw.

3.1.4 System performance

The initial results reported in this thesis on the work towards buffer-gas load-ing of atomic chromium have all been obtained with the original cryostat. In its full configuration, with both the magnet and the experimental cell installed, the dilution unit has a measured cooling power of 100W at 100 mK and is able to reach a base temperature of about 21 mK. It hence operates within the required specifications and should be suitable for an actual buffer-gas load-ing experiment, which is also supported by the first experimental experience (Section 5.4).

The amount of liquid helium consumption, which is largely independent of the operating state of the dilution unit, is illustrated in Figure 3.3. Under stan-dard conditions, the helium level in the upper dewar declines with an average rate of 4.5 cm/h, yielding a hold time of only 11 h. By removing the current leads for the magnet, it can be increased to about 26 h, indicating that their vapor cooling has not been optimal. Both numbers however do not allow a reasonable continuous operation of the cryostat.

3.1.5 Retrofit of the LHe dewar

The situation has improved with a major upgrade of the liquid helium dewars.

First evaporation data from the retrofitted cryostat indicates an increase of the hold time by a factor of two (Figure 3.3), which now is approaching the required specifications.

The relevant changes in layout are summarized in Figure 3.4. 11 vapor-cooled radiation baffles and a thinner wall of the upper dewar significantly reduce the heat load on the LHe bath and lower the helium consumption to a tolerable amount. Additionally, the IVC is now completely surrounded by LHe, which increases the storage volume and at the same time avoids radiative heating of the IVC. The 1 K pot fill line has been extended, so that the dilution unit can keep running even when the upper helium tank is empty.

SUPERCONDUCTING MAGNET 41

0 10 20 30 40

0 10 20 30 40 50 60 70

Levelinupperdewar[cm]

Time [h]

new design, no leads

old design, full config.

old design, no leads

Figure 3.3:Measured helium consumption in the upper dewar. The dashed line shows the decline of the LHe level for the fully equipped original cryostat. When the vapor-cooled current leads are not installed, the asymptotic evaporation rate reduces to a value of 1.5 cm/h. With the new design, this number is improved to 1.0 cm/h at even higher LHe levels.

Further improvements of the hold time should be expected with an opti-mization of the gas flow through the vapor-cooled current leads for the magnet.

This will be accomplished by installing needle-valves to carefully balance the impedance of the standard recovery line against that of the current leads.