We report advances in the adiabatic calorimetry technique that permit a determination of heat capacity to at least 100ppm. The resolution is ultimately limited not only by the thermometry, but also by how well we can control the heat input to the cell.
Bibliographical noteFunding Information:
the flux transformer wires should be broken in two to improve the thermal isolation. Shield integrity can be protected by an overlapping NbTi sleeve. The strongest thermal link between our cell and the main stage is now due to 3 hollow Vespel support rods. Scatter in a heat capacity measurement is also limited by how well one can measure the heat pulses and how well one can control the fluctuations in the heat leak to the cell. We use rectangular heat pulses and measure the voltage across a room temperature 4ppm/K current sense resistor. Every wire or fill line used is extensively heat sunk at the main stage, glued or soldered over a 20-50cm length, before being coiled up and attached to the cell. The cell is enclosed by a radiation shield that is thermally anchored at this main stage. The main stage is regulated by a germanium resistance thermometer and the regulator setpoint is iteratively adjusted until the drift in the cell temperature, generally less than O.lnK/sec, is minimized. Our experiment is designed to work a few mK to a few 1OOmK below the X-point, where the cusp singularity in 4He-filled aerogel has been observed (see Figure 2). The total heat capacity is huge, over lJ/K for our 19Dx2.6H mm sample, resulting in an external thermal isolation time, between the cell and the main stage, of 2.6 x 105sec or 3.0days. The internal equilibration time, within the cell itself, is about 30sec. We fit the temperature versus time data to a common linear drift on both sides of the pulse, plus a quasi-exponential decay after the pulse. Deviations from exponential decay appear in the 5th digit because the external isolation time is temperature dependent. We avoid a recursion problem by modeling the isolation time as a linearly time dependent quantity, which is reasonable because the temperature change is very nearly linear. We acknowledge the support of the National Science Foundation through grant NSF-DMR84-18605 and the Cornell Material Sciences Center through grant NSF-DMR85-16616-A02. G.K.S.W. acknowledges the sup-
port of an A.D. White Fellowship from Cornell University and a 1967 Postgraduate Fellowship from the Natural Sciences and Engineering Research Council of Canada. P.A.C. acknowledges the support of an AT&T Bell Laboratories Ph.D. Scholarship.