Abstract
Precise heat-capacity results are presented for adsorbed on graphite. The temperature range of the data is from 2 to 200 mK, while the coverages span from somewhat below monolayer completion up through five atomic layers. Promotion of atoms into the second, third, and fourth layers is clearly observed. Nuclear-spin exchange energies of the order of a few tenths of a mK are found for the submonolayer incommensurate solid phase. These values differ significantly from those recently inferred from NMR experiments. Data for the second-layer fluid yield quasiparticle effective masses that agree well with the corresponding first-layer values and range from one to five times the bare mass. Prior to third-layer promotion, the second layer undergoes a first-order phase transition. By comparison with the phase diagram for the first layer, the new phase in the second layer is assumed to be a registered solid. Registry is now with respect to the first layer, which continues to exist as a triangular-lattice solid incommensurate with the graphite substrate. The registered phase exhibits a large, sharp heat-capacity anomaly at 2.5 mK.
This anomaly may be due to antiferromagnetic polarons which form around zero-point vacancies or may be the signature of an unusual registered phase in which some of the atoms are positioned at substrate potential maxima. As the coverage is increased further, the second-layer spin peak remains located at 2.5 mK but suddenly grows in amplitude, while the temperature dependence above the peak changes from towards . The anomaly reaches its greatest magnitude at 0.24 atoms/A where, perhaps coincidentally, promotion of atoms into the fourth layer also occurs. At this same coverage previous magnetization measurements have shown a large ferromagnetic peak. The heat-capacity data indicate that the ferromagnetic peak occurs when the second layer exists in a state intermediate between a registered solid and the incommensurate solid. Consistent with this observation, the spin system at 0.24 atoms/A cannot be accurately described by a nearest-neighbor Heisenberg Hamiltonian. This is contrary to the situation at somewhat higher coverages and to the finding from recent magnetization experiments.
- Received 14 September 1989
DOI:https://doi.org/10.1103/PhysRevB.41.1842
©1990 American Physical Society