The New SMC Dilution Refrigerator

Transcription

1 The New SMC Dilution Refrigerator J. M. Kyynäräinen a, M.P. Berglund b, and T.O. Niinikoski a a CERN, 1211 Geneva 23, Switzerland b Low Temperature Laboratory and Institute of Particle Physics Technology Helsinki University of Technology, Espoo, Finland Abstract. The dilution refrigerator for the SMC twin solid polarized target at CERN is by far the largest one in operation. The 2.5 liter, 1.5 m long target is loaded horizontally into the cold mixing chamber. A cooling power of 400 mw at 300 mk temperature is available for dynamic nuclear polarization. Proton and deuteron targets are polarized to ±94% and ±47%, respectively. The base temperature of 30 mk enables rotation of the polarization vector without losses. INTRODUCTION The experiment NA47 of the Spin Muon Collaboration (SMC) at CERN SPS determines the spin dependent structure functions of the nucleons by scattering polarized muons from solid polarized proton and deuteron targets (1,2). The target material is either normal or deuterated butanol frozen into glassy beads. The nuclear spins are polarized with dynamic nuclear polarization (DNP), based on microwave saturation of impurity electron spins near their paramagnetic resonance in a 2.5 T longitudinal field. In our experiment the same beam penetrates oppositely polarized target halves to minimize the effect of beam flux variations. The spin directions are frequently reversed to cancel residual false asymmetries. This is done by rotating the magnetic field at 0.5 T. Dilution refrigeration is essential because of the possibility to effectively "freeze" the spins as the spin-lattice relaxation becomes very slow at low temperatures and thereby allows the field rotation procedure with minor loss of polarization. CONSTRUCTION OF THE REFRIGERATOR The geometry of the refrigerator is dictated by the requirement to provide minimum amount of additional material in the muon beam path and by the handling of the target material below 100 K. In Fig. 1 the refrigerator and the magnets (3) are shown with the target in place.

2 VACUUM VESSELS SUPERINSULATION He-3 PRECOOLER He-4 EVAPORATOR SEPARATOR AND SIPHON THERMAL RADIATION SHIELDS DILUTION REFRIGERATOR MICROWAVE CAVITIES TARGET HOLDER TARGET MAGNETS ν ν 1 m ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν ν 7. 5 FIGURE 1. The longitudinal cross-section of the dilution refrigerator, shown together with the magnets. The muon beam enters the cryostat from the left. µ The Precooler, the Evaporator and the Still Compared with refrigerator (4) of the predecessor experiment NA2 of EMC, our refrigerator has a considerably more elaborate system for precooling. A schematic flow diagram of the refrigerator, showing the arrangement of the heat exchangers, is shown in Fig He vapour at atmospheric pressure is taken from a phase separator to precool the incoming 3 He in a parallel tube counterflow heat exchanger (No. 1 in Fig. 2). This is also in thermal contact with the low-pressure 3 He and 4 He in a 4-flow heat exchanger (No. 2). Fin type exchangers have been chosen to give reasonable pressure drop with high gas flow at low pressure. The evaporator volume is 27 liters, which gives ample space for a 3 He condenser and provides good tolerance to sudden heat load variation without risk of drying up or overflowing. A copper grid type heat exchanger (No. 3) is used to cool the incoming 3 He and the 4 He liquid from the separator. This reduces the 4 He boil-off rate by up to 30%. During initial cooldown the heat exchanger can be bypassed. The condenser (No. 4) is made of copper tubes with inner surface area of 0.35 m 2. Between the condenser and the Joule-Thompson needle valve a fin/grid type heat exchanger is placed in the still pumping tube (No. 5). The still has two main volumes, the lower one containing a tubular heat exchanger for incoming 3 He and the upper one a heater and a level gauge. The heat exchanger (No. 6), made of copper tubing, has surface areas of 0.12 m 2 and

3 0.23 m 2 for the 3 He-rich and the dilute phases, respectively. The still heater is made of a stainless steel strip, placed vertically and having a surface area of 0.58 m 2. At the maximum design flow of 500 mmol/s about 16 W of heating power is needed, corresponding to a power density less than 3 mw/cm 2 to avoid film boiling. No special precautions were taken to suppress the superfluid film creep into the still pumping tube, as the estimated film flow rate is much smaller than the lowest practical 3 He flow rate. HE-3 PUMP HE-4 PUMP HE-4 RECOVERY HE-4 DEWAR SEPA- RATOR EVAP- ORA- TOR 5 STILL MIXING CHAMBER TARGET TARGET FIGURE 2. The arrangement of the heat exchangers in the refrigerator. The Main Heat Exchanger Between the still and the mixing chamber a tubular (No. 7) and a sintered copper (No. 8) heat exchanger are placed into a helicoidal groove formed by glassfibre-epoxy spacers. The groove defines also the dilute phase flow channel. A machined stainless steel tube with tight fitting to the spacers seals off the flow channel. The length of the flow channel along the centerline is 1.7 m and the free cross-sectional area is about 11.5 cm 2. The latter was calculated using data from previous smaller dilution refrigerators (5). The tubular heat exchanger is used to condense vapour bubbles which may have formed in the flow at the expansion needle valve and to cool the helium below the phase separation temperature before entering the narrow channels of the sintered heat exchanger. The unit is made of flattened stainless steel tubes and has a total inner surface area of 0.1 m 2. The sintered heat exchanger consists of 12 elements with dimensions of 12 x 4 cm 2 arranged in two parallel streams. To prevent cold plug formation due to the increasing viscocity of 3 He below 0.5 K, the flow in the streams is crossed at several points. The base material is 0.2 mm Cu(P) foil. The elements consist of two sintered plates which were first bent to fit into the flow channel and then electron beam welded together. The average thickness of the layer of the

4 nominally 18 µm grain size sinter is 0.75 mm, yielding 375 g of sinter and a geometrical surface area of 12 m 2 on both the concentrated and the dilute phase streams. The heat transfer is enhanced by grooves on the sinter surface to increase the interfacial area with fluid flow and to produce turbulence in it. The calculation of the required surface area was based on the principles of optimized design of Ref. 6. As the refrigerator has to provide high cooling powers at high temperatures, the effective surface area cannot be increased by increasing the sinter thickness, as the transverse heat conduction in the helium and in the sinter limits the heat transfer. Instead, one has to rely on increasing the interfacial surface area between the sinter and the fluid streams. The Mixing Chamber, the Target Holder and the Microwave Cavity The mixing chamber, with a length of 1500 mm and a diameter of 70 mm, is made of glassfibre reinforced epoxy with 0.6 mm wall thickness to ensure sufficient rigidity and to withstand slight overpressures in the case of a pump failure. The target holder slides into the horizontal access tube and is sealed with a cold indium seal at the still back flange. A stainless steel vacuum chamber provides thermal isolation and a plastic part confines and supports the material. The vacuum chamber has two 0.1 mm stainless steel windows for the beam access and 6 aluminium foil thermal shields, and it provides thermal anchors for the coaxial lines for NMR and for the instrumentation wires. The plastic part is mostly made of Kevlar-epoxy composite for rigidity and reduced thermal contraction. The target material is located in two cells with a length of 650 mm and a diameter of 50 mm, separated by 200 mm. The net volume of the target is 2x800 cm 3. Good heat transfer is assured by making the cells of polyester net with 60% open area. The containers weigh only 30 g each. To ensure uniform temperature, 3 He is fed into the mixing chamber through 40 holes in a CuNi tube which is fixed to the target holder. A spring-loaded conical connector couples the outlet of the heat exchanger to the CuNi tube when the target holder is in place. The mixing chamber is surrounded by a cylindrical microwave cavity of 210 mm diameter, made of copper. The cavity is divided axially in two compartments by placing graphite foil coated copper baffles and copper reflectors in the center. Inside the mixing chamber isolation was obtained using graphite coated Nomex honeycomb absorbers and fine copper mesh reflectors, designed to ensure free diffusion and convection in the dilute solution. The microwave isolation was measured to be db in the GHz band with empty cavities; this prevents the microwave power from leaking excessively from one side of the cavity to the other. The cavity is cooled to 3 K by 4 He flow controlled by a cold needle valve. The two wave guides feeding the cavity enter the main vacuum through FEP windows and continue with circular sections made of silvered thinwalled CuNi tubing and further with rectangular ones. FEP windows close the

5 guides before the coupling slots to the cavity, isolating its vacuum from the main vacuum and preventing the loss of 3 He in case of mixing chamber rupture. PUMPS We use the same pumping system for 3 He as the EMC experiment with 8 Root's blowers in series, giving a volume speed of m 3 /h. This is sufficient to lower the pressure of the still to 0.2 mbar at the flow rate of 30 mmol/s and 1.2 mbar at 200 mmol/s. The pumps are connected to the refrigerator via a 20 m long line, 320 mm in diameter. At a flow rate of 200 mmol/s the pressure drop is about 0.4 mbar across both the 320 mm line and the heat exchangers above the still. Charcoal traps at room temperature and at 77 K are used to filter out impurities in the 3 He return flow, in addition to the zeolite filters in the pumps. The 4 He evaporator is pumped by a Root's blower followed by a rotary pump with a nominal speed of 2000 m 3 /h, maintaining the evaporator at 1.5 K. The cavity cooling line is connected directly to the rotary pump. CONTROL, DATA ACQUISITION AND SAFETY SYSTEMS Temperature measurement below about 10 K is based on carbon and RuO resistors, read by 4-wire AC resistance bridges. Two carbon resistors are dedicated to monitor the microwave power at the target; other resistors are partly shielded against the microwave field. Higher temperatures are measured using Si diodes. High-accuracy pressure measurements, e.g. that of the 3 He vapour pressure during NMR signal calibration, are done using capacitive pressure gauges. The evaporator and still levels are monitored by capacitance bridges. CRYOSTAT VACUUM SYSTEM NMR SYSTEM CONTROL INSTRU- MENTS MEASURING INSTRU- MENTS GPIB MICRO- WAVE SYSTEM INTER- LOCKS DATA LOGGER/ PLOTTER SL-GMS GUI ETHERNET UNIX WS ERROR MESSAGES RS232 CPU DAC SIGNAL CONDI- TIONER ADC GPIB INTER- FACE VT100 VME BUS FIGURE 3. Block diagram of the cryogenic control, data acquisition and safety system.

6 A graphical user interface, based on a commercial software package SL-GMS, is running in a Unix workstation and X terminals. The control programs are running in VME processors. More than 100 cryogenic parameters are logged by the programs which also generate alarms. Most readout instruments are read via a GPIB bus, either directly or using a 32-channel data logger/plotter (see Fig. 3). The controlled parameters include the needle valve for evaporator filling, the separator and microwave cavity flow rates and the still heater. The latter is to be controlled by the microwave power to maximize the cooling power. A PLC-based interlock system, powered by an uninterruptible supply, protects the target against loss of polarization and of 3 He. PERFORMANCE The cooldown of the magnet from room temperature to 4 K takes about one week. The dilution refrigerator is then cooled in 8 hours to 77K, the cells of the target holder are filled with the material in a special LN 2 bath and the target holder is pushed into the precooled mixing chamber. Purging, final cooldown and condensing the 4 He- 3 He mixture (8700 litres STP) takes less than 8 hours. During DNP the temperature of the helium mixture decreases slowly from about 350 mk to 200 mk as the optimum microwave power is reduced with increasing polarization. The maximum polarizations obtained in the proton and deuteron targets are ±94% and ±47%, respectively. About 95% of the maximum polarization is reached in 12 hours of DNP. The target is cooled down below 100 mk by turning the microwave power off hours before the field rotation. The ultimate temperature is reached only after several hours of running without microwaves. This suggests that the microwaves heat the isolator and that the thermal contact is sufficiently poor to the mixing chamber to prevent a rapid cooldown. However, practically no loss of polarization takes place during the field rotation. The cooling power in the mixing chamber with optimum 3 He circulation is shown in Fig. 4. The temperature was measured at the outlet of the dilute phase from the mixing chamber. Residual heat leak of about 1 mw to the mixing chamber is mainly from this end, and the temperature there is mk higher than in the downstrean end where the lowest temperature of 30 mk is obtained. The 3 He flow rate has practical minimum and maximum values of 27 and 350 mmol/s with the 4 He contamination of about 25% which is a consequence of the rather high still temperature of K. The liquid 4 He consumption of the refrigerator varies between 15 and 40 l/h depending on the 3 He flow rate. This compares favourably with the maximum of 70 l/h of the EMC dilution refrigerator, and gives a safety margin to the helium liquifier with a maximum capacity of 100 l/h and which supplies also the magnets.

7 Maximum cooling power [mw] Mixing chamber temperature [K] FIGURE 4. The maximum cooling power vs. mixing chamber temperature. CONCLUSION Dilution refrigeration has been shown to be a very suitable method to reach high polarizations in a large target and to manipulate the polarization vector. The practical limit of the target size is set by the cooling power of the refrigerator, determined by the available pumping speed of 3 He. Several innovations have improved the reliability considerably, e.g. the glassfibre-epoxy mixing chamber, vacuum brazed heat exchangers and the use of demountable metal seals at superfluid 4 He temperatures. Computer control and graphical user interface enable nonexperts to handle the refrigerator on routine basis. ACKNOWLEDGEMENTS We wish to thank G. Bonnefond, M. Bron, R. van Danzig, B. Feral, D. Geiss, H. Herbert, J. Homma, M. Houlmann, A. Isomäki, J. Kaasinen, S. Kaivola, J.-C. Labbe, Y. Lefevere, L. Naumann, A. Staude, B. Trincat, and S. Utriainen for their help in machining, assembling and testing the parts of the refrigerator. REFERENCES 1. B. Adeva et al., Phys. Lett. B 302, 533 (1993). 2. D. Adams et al., Phys. Lett. B 329, 399 (1994). 3. A. Daël et al., IEEE Trans. on Magnetics 28, 560 (1992). 4. T. O. Niinikoski, Nucl. Instr. and Meth. 192, 151 (1982). 5. T.O. Niinikoski, Nucl. Instr. and Meth. 97, 95 (1971). 6. T.O. Niinikoski, "Dilution refrigeration: new concepts", in Proceedings of the Sixth International Cryogenics Engineering Conference, 1976, pp