The whole probe

For the measurements on Sr2RuO4, low temperatures (as low as sub-Kelvin) and

magnetic field were both needed. Therefore I used a dilution refrigerator (Kelvi- nox 25 from Oxford Instruments), whose base temperature was 50 mK, to generate the low temperature, and a three dimensional (3D) vector magnet (1T◊1T◊9T) to provide the magnetic field. A short introduction to the refrigeration method will be given later. Now let me introduce first the probe that was needed for the measurements.

During the measurements, the sample should be placed right in the field centre, and at the same time good thermalisation was also important. A probe was designed to fulfil these requirements, which was composed of the strain device, a copper frame, a hollow stainless steel tube, a copper plate and some additional small components. One end of the stainless steel tube was inserted in to the central hole of the copper plate, whose size matched the outer diameter of the tube, and the two were glued together with Stycast 2850FT. The other end of the tube was inserted into a whole of the copper frame, whose dimension was just big enough for the tube, and the two were also then glued with Stycast 2850FT. So the copper frame, the stainless steel tube and the copper plate were glued together, with the stainless steel tube in the middle; they together will be called the frame of the probe. The copper plate had several other holes through which it could be fixed

against the cold plate of the dilution refrigerator with screws. The length of the stainless steel tube was chosen such that when the stain device (with the sample mounted) was installed into the copper frame the sample would be situated in the field centre.

Figure 3.5: The real strain device used for the measurements on Sr2RuO4,

installed into the copper frame of the probe. The strain device was fixed by two screws which could be seen on the top and bottom sides of the frame. The mounted sample is hidden underneath the coils for magnetic susceptibil- ity measurements. A thermometer was installed underneath the thermometer protection structure. The sample plates and the test sample were electrically isolated from the rest of the strain device by two pieces of cigarette paper, as

shown around the screws in the middle.

The copper frame was designed to accommodate the strain device and to fix it stably throughout the measurements. The main body of the strain device had a threaded hole (M2) on each side, and the copper frame had two corresponding holes which could let M2 screws go through (see Figure3.5). To avoid too much lateral deformation of the stain device caused by the screws, they should not be screwed too tightly, however on the other hand, considering the thermal contraction, which was different for copper and titanium, the screws should be screwed relatively tightly because otherwise it would become too loose as the whole probe was cooled down. To achieve the right compromise, I used a crinkle washer for each screw to adjust the tightness, and the result turned out to be satisfactory.

In order to enable temperature sweeps, a heater should be installed on the probe and be thermally connected to the sample. The most convenient option for a heater was a resistor, and ideally a film resistor because bigger contact area means better heat conductance. However the strain device was too limited in size for a film resistor to be installed on, thus eventually I glued a 1000 film resistor onto the copper frame with GE varnish. Then I used a piece of silver foil to enhance the thermal conductance between the copper frame and the strain device. A small thermometer was installed on the strain device (beneath the thermometer protection structure in Figure 3.5) to measure the temperature of the sample. To reduce the errors brought by the thermal time constant between the sample and the thermometer, I placed a piece of silver foil underneath both the thermometer and the sample plates. As we shall see in Chapter 4, the hysteresis in temperature up runs and temperature down runs was very small, which proved the validity of this method.

A 24-way connector and 12 pairs of twisted CuBe wires were used to make elec- trical connections from the bottom of the probe to the 24-way connector of the dilution refrigerator beside the mixing chamber, which was then linked to the measurement instruments in the laboratory. These wires were what were used to make connections for the thermometer, for the heater and for the coils for mag- netic susceptibility measurements. CuBe wires were chosen because of their high thermal resistance.

Two heat sinks (not seen in Figure 3.5) were installed on the copper frame in order to thermalise the measurement wires and coaxial cables. The thermal time constant between the cold plate of the dilution refrigerator and the copper frame was found to be too long for the measurements, therefore I added a silver wire (0.5mm in diameter) to reduce it and the outcome turned to be good. One more thing to stress is that to drive the piezo-stacks at low temperatures, the maximum voltage needed was several hundred Volts, high enough to cause potential danger. Therefore I used a pair of coaxial cables (the same type as that used for capacitance measurements), for which the maximum operating voltage was 10 kV, to make electric connection between the stacks and the output terminals of a bi-polar high

voltage amplifier, that was home designed to drive the piezo-stacks. The high voltage amplifier could provide static voltages between -1000 V and +1000 V, with gain from 1 and 100. Even though coaxial cables were used for safety, all the connecting points (e.g. where connectors were used) were still carefully dealt with at all times during the operation, and the output voltages of the high voltage amplifiers monitored constantly.