Can a topological insulator turn heat into spin current?
A thin film of the topological insulator bismuth selenide turns a temperature gradient into a flow of electron spin far more efficiently than platinum or tungsten.
Source
Thermally generated spin current in the topological insulator Bi<sub>2</sub>Se<sub>3</sub>
Study at a glance
- Design
- Other — Thin-film Bi2Se3/CoFeB Hall-bar devices with on-chip heaters; compares thermally driven (spin Nernst) and electrically driven (spin Hall magnetoresistance) signals, plus second-harmonic Hall torque measurements.
- N
- No participant count; main bilayer device plus a second bilayer with 6 nm CoFeB and single-layer control samples.
- Population
- MBE-grown 8 nm Bi2Se3 films capped with sputtered CoFeB, measured at room temperature
- Outcome
- Spin Nernst magneto-thermopower, spin Hall magnetoresistance, Seebeck coefficient, and derived spin Nernst ratio
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
The authors grew 8 nm films of bismuth selenide, confirmed their surface states, and topped them with a thin magnetic CoFeB layer patterned into long Hall bars with small heaters at each end. They measured how the bar's resistance and heat-driven voltage changed as a strong magnetic field was rotated, separating the spin-related parts from ordinary magnetoresistance by their field dependence and by control samples of each layer alone. Combining these with a measured Seebeck coefficient and a spin-torque estimate, they calculated how efficiently heat generates spin current.
What they found
The heat-driven signal had the angular pattern expected for the spin Nernst effect and flipped sign when the gradient was reversed. The ratio of the spin Nernst to spin Hall efficiency was about −0.83, and the spin Nernst ratio had a lower bound of about −0.62 in magnitude, roughly three times that of platinum and two to three times that of tungsten. Spin current per unit thermal gradient was higher than in tungsten and similar to platinum despite bismuth selenide's much higher resistivity.
The limits
What it doesn't show
The key efficiencies are lower bounds, because the spin Hall ratio was estimated from a torque measurement that includes an unknown interface transparency factor. The analysis assumes heat-driven and electrically driven spin currents are converted to voltage the same way. The study used one composition at room temperature, so the proposed tuning of the Fermi level is untested, and it cannot tell whether topological surface states survive under the magnetic layer or are responsible for the effect.
Key terms
- Spin current
- A flow of spin angular momentum, which can occur even without a net flow of electric charge.
- Spin Nernst effect
- Generation of a transverse spin current by a temperature gradient in a material with strong spin-orbit coupling.
- Spin Hall magnetoresistance
- A change in a bilayer's resistance with magnetization direction caused by spin current reflecting from a magnetic interface and converting back into charge current.
- Seebeck coefficient
- The voltage produced per unit temperature difference across a material in open circuit.
- Mott relation
- A link between thermoelectric coefficients and the energy derivative of the corresponding electrical conductivity at the Fermi level.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What effect converts an in-plane temperature gradient into a vertical spin current?
Common questions
How can a voltage measurement detect a spin current?
Spin current reflecting from the magnet depends on the magnetization angle and is converted back into a voltage by the inverse spin Hall effect, so the voltage varies with the square of the cosine of the field angle.
Why is the spin Nernst ratio only a lower bound?
It depends on the spin Hall ratio, which was estimated from spin-torque efficiency; since interface transparency is at most one, the true spin Hall ratio and thus the spin Nernst ratio could be larger.
How might the effect be made even stronger?
The Mott relation says it depends on how steeply the spin Hall conductivity changes with energy, so tuning the Fermi level by doping could increase it.
More on Topological materials