Anisotropy of the magnetic susceptibility of UPt₃

Despite great efforts, both experimental and theoretical, the problem of magnetic behavior of UPt₃ and its relation to the multiple superconducting phases has not been completely solved. This system continues to show unexpected and often unexplained features. To study the magnetic anisotropy of UPt₃ single crystals in a wide range of temperatures and magnetic fields, we used a capacitive torque-meter, where the torque acting on the probe is proportional to (χ_a,b - χ_c)B². As in the old data of Frings and Franse we find in intermediate fields that χ_a and χ_b exhibit a maximum around 20 K, while χ_c is only slightly enhanced with respect to the room temperature value. The resulting torque maximum around 20 K, however, turns out to be strongly field dependent. On decrease of B, a shift towards lower temperatures sets in below 3 T. This shift becomes nearly horizontal between 0.5 and 0.3 T and the maximum moves down to below 10 K. A second unexpected feature occurred when we followed the temperature dependence of χ_a,b - χ_c down to 15 mK: it ends suddenly but reversibly at a well-defined border line L_i which is also field dependent. For magnetic fields above B_c2(0), L_i lies between 60 and 100 mK and joins the upper critical field curve at (100 mK, 1.86 T). Below this point it follows B_c2(T) down to the tetracritical point (T*, B*) where it splits off again. These results will be discussed with respect to existing theories of the superconducting order parameter and its relation to the anisotropic magnetism of UPt₃.