Control of the low temperature resistance plateau in the topological Kondo insulator SmB6
Phys. Rev. X 4, 031012 (2014)
Accepted Phys. Rev. Lett. (2014)


Executive Summary

A topological insulator is a newly (c.a. 2006) discovered type of material that is electrically insulating in the bulk but has metallic states on the surface at interfaces with normal insulators (such as air, vacuum, most materials, etc.).


The presence of a metallic surface is by itself a remarkable property that has numerous potential practical applications. But the surface states of a topological insulator have an even more amazing feature: spin-momentum locking. In a normal metal, the direction of electron motion is independent of the spin of the electron. Not so for topologically protected surface states: the direction of electron motion is inextricably linked to the orientation of the spin. This allows for "automatic" generation of spin-polarized currents for spintronics, new kinds of quantum computation, and possible realization of new electronic states of matter (quasiparticles that obey different statistics than all other known particles).


Experimentally known topological insulators are based on a limited range of elements and structure types. Samarium hexaboride, SmB6, is a material that undergoes a transition from a metallic to an insulating state below ~ 40 K. At even lower temperatures, ~ 5-10 K, a resistance plateau indicative of residual metallicity appears. The origins of this remnant conduction have remained enigmatic for more than 40 years. Previously attributed to impurities and secondary phases, recent theory and experiments suggest that it is topologically protected metallic surface states, arising from a non-trivial band topology in the insulating state of SmB6, that underpin the low temperature resistivity plateau.


We show a correlation between thermodynamic observables (e.g., specific heat) and the low temperature electrical conduction in SmB6. Further, we elucidate the role of two common impurities - aluminium inclusions from flux growth and ever-present carbon - on the physical properties. We find that the surface conductivity of our samples increases with bulk carbon content, suggesting that resistance can be controlled via the addition of carbon. In contrast, aluminum does not dramatically affect the resistance plateau.


When combined with neutron scattering and X-ray absorption studies, we conclude that the actual story is rather more complex. Neutron scattering provides strong evidence that SmB6 is a topological insulator. However, X-ray absorption measurements directly show that the surfaces of SmB6 have a different formal valence for Sm than found in the bulk, suggesting that non-topological metallic surface states exist due to surface reconstructions. The most plausible way to reconcile these measurements with the carbon and aluminum studies is that SmB6 is a topological insulator, but that the low temperature resistance plateau originates from both trivial and topological metallic surface states.


These results serve as a foundation for resolving the origins of, and controlling, the surface states in SmB6, a material appealing for spintronics and quantum computation applications. Further, some of the observed behavior - a 5d rather than 4f form factor in the neutron scattering and a huge entropy loss during the formation of the inuslating state - are seemingly at odds with the traditional Kondo picture and deserve proper theoretical treatment.