MEASUREMENT OF MICROKELVIN TEMPERATURE DIFFERENCES IN A CRITICAL-POINT THERMOSTAT

The density of a pure fluid near its critical point is extremely sensi tive to temperature gradients. In the absence of gravity, this effect limits the fluid's homogeneity. For example, at 0.6 mK above the criti cal temperature, the microgravity experiment Critical Viscosity of Xen on (CVX) can allow temperature differences no larger than 0.2 mu K, co rresponding to a gradient of 10(-5) K.m(-1). The CVX thermostat, which consists of a thick-walled copper cell contained within three concent ric aluminum shells, was designed to achieve such a small temperature gradient. However, asymmetries not included in the thermostat's model could degrade the thermostat's performance. Therefore we measured the temperature gradient directly with a miniature commercial thermoelectr ic cooler consisting of 66 semiconductor thermocouples. We checked the results with a half bridge consisting of two matched thermistors. The measurement was made along a thin-walled stainless-steel cell whose c onductance was much lower than that of the copper cell, thus ''amplify ing'' the temperature differences by a factor of 60. When the thermost at was controlled at a constant temperature, the steel cell's static t emperature difference was 5 +/- 1 mu K. (The value inferred for the co pper cell is 0.08 mu K.) Ramping the thermostat's temperature at a rat e of 1 x 10(-5) K.s(-1) increased the temperature difference to 0.36 m K. These results demonstrate the feasibility of achieving extremely lo w temperature gradients.