The neutron scattering function for liquid 3 He at 120 mK and pressures of 0, 0.5, 1, and 2 MPa has been determined for wave vectors in the range 3 nm -1 < Q < 20 nm _1 . This represents the first experimental information on the density dependence of the zero-sound frequency and damping at finite wave vectors, and should serve as a useful test of theories on excitations in neutral Fermi liquids. The results are in qualitative agreement with extensions of the Landau Fermi-liquid theory to finite wave vectors.PACS numbers: 67.50.Dg Neutron inelastic scattering offers a unique method to examine directly the elementary excitations in liquid 3 He on a microscopic scale, i.e., for wave vectors of order 10 nm -1 . Through their spin-dependent interaction with the 3 He nuclei, neutrons probe the nuclear spin fluctuations, in addition to the density fluctuations which are probed by the spin-averaged interaction. In the neutron scattering function, one distinguishes three components: the excitation of single particle-hole (p-h) pairs, a collective density mode (zero sound), and a broad distribution corresponding to multiple p-h excitations. The current theoretical and experimental situation in this area has recently been reviewed by Glyde. l Previous neutron-scattering experiments 2 " 5 have shown that the single p-h spectrum resembles the one expected for a noninteracting Fermi gas but with an effective mass m* ~ 3mo which lowers the energy of the p-h band compared with that of the noninteracting system. For Q < 10 nm -1 , the single-pair spin fluctuations are enhanced and are well described by the paramagnon model. 6 The zero-sound mode is well resolved for wave vectors less than 10 nm -1 and shows anomalous dispersion, i.e., the energies are above those corresponding to linear dispersion co^coQ, where Co is the ultrasonic zero-sound velocity. It broadens considerably with increasing wave vector even before the mode overlaps with the p-h band. All the results published so far at saturated-vapor pressure (SVP) are in good general agreement with the predictions of the polarization-potential theory developed by Pines and co-workers. 7 Here the potential parameters, obtained by a fitting to the neutron-scattering results and to the Landau parameters, have been used to calculate transport properties for both normal and superfluid 3 He which are in excellent agreement with measured values. 8 The effective interactions in the polarization-potential theory are expected to be sensitive to the density, as discussed by Aldrich and Pines 7 and as shown by the Landau parameters, which are the low-(? limits of the effective interactions.While in an earlier experiment 9 the scattering function for pressurized liquid 3 He could only be measured for wave vectors larger than 12 nm -1 , the present experiment explores the wave-vector range 3 nm ~l < Q < 20 nm" 1 . This therefore represents the first measurement of the zero-sound mode at elevated pressures.The measurements were made at the time-of-flight spectrometer 10 IN6 at the high-flux...

No abstract

We have measured and calculated the density and spin-density dynamic structure factors S c (Q, ) and S I (Q, ) of normal liquid 3 He as a function of wave vector Q and temperature T. The static spin susceptibility (T) and specific heat C V (T) are also calculated. These properties all depend upon the effective mass m*(k, ) of the Fermi quasiparticles making up the liquid. We use a model in which m* peaks near the Fermi surface to m*ϭ2.8, the Landau theory effective mass, and decreases toward the bare mass m*ϭ1 for quasiparticles away from the Fermi energy ⑀ F . The theory for all the properties may be viewed as Landau theory with an effective mass m*(⑀ k )ϭm*(k) that decreases as the quasiparticle energy ⑀ k moves away from ⑀ F . The peaking of m* at ⑀ F is widely predicted in Fermi systems and the aim is to test how important this physical feature is in the dynamics of liquid 3 He. We find that S c (Q, ) and S I (Q, ) versus Q and T as well as (T) are well reproduced by the model for the same m*(k). The C V (T) can be reproduced, but a much lower value of m*(k) at energies ⑀ k away from ⑀ F is required, m*Ӎ0.5, as found in previous calculations of C V (T). We conclude that the peaking of m* at ⑀ F is an important physical feature to include in calculations of S (Q, ) and that the quasiparticle model itself is inadequate for C V at higher temperatures.

No abstract

No abstract