A Bayesian Approach to Nuclear Spin Relaxation Phenomena in Topological Kondo Insulator SmB 6

Abstract

In studies of the promising topological Kondo insulator (TKI) candidate SmB 6 , nuclear magnetic resonance (NMR) measurements of the spin–lattice relaxation rate, 1/T 1 , have provided crucial insights into this compound: for example, the unique opening of a small hybridization gap within a narrow band at low temperatures. However, the origin of the hump-like behavior of 1/T 1 reported below 20 K remains poorly understood. To address the mechanisms of low-temperature T 1 relaxation, we have precisely analyzed the 11 B-NMR T 1 relaxation curves of single-crystalline SmB 6 by introducing a Bayesian inference framework, which provides insights beyond the conventional least-squares fit in terms of a quantitative measure of fitting quality. Our comparative analysis using three different models reveals that the T 1 relaxation curve, which is well described by a single component above ∼100 K, evolves into a multi-component regime below ∼60 K, suggesting the inhomogeneous development of the hybridization gap. Below ∼20 K, 1/T 1 in the present study also exhibits an anomalous enhancement, which remains clear even after considering the distribution of T 1 components. The analysis suggests that this low-temperature enhancement is an intrinsic property of this compound, and that a novel relaxation mechanism — distinct from the high-temperature process driven by the density of states — emerges in this temperature region. These observed phenomena are discussed based on the analogy with other potential TKIs, SmS, and YbB 12 .