Dilute bismides make thermoelectric debut
A US academic is about to embark on a program that he believes will lead him to the most efficient thermoelectric materials ever, with the help of dilute bismides.
University of Delaware s Joshua Zide has earned a 3-year, $510,000, grant from the US Office of Naval Research (ONR) to investigate the unusual semiconductors and nanocomposites.
“Using these materials for thermoelectrics is to the best of my knowledge fundamentally new and no-one else has proposed it,” Zide told compoundsemiconductor.net.
Having won the award in April, Zide has bought the necessary source materials and effusion cells for his MBE system and plans to start growing material “within weeks”.
The ONR is sponsoring Zide s work in order to harvest the waste heat produced by power generation systems on its vessels. Solid-state thermoelectric devices could be particularly important on a nuclear submarine, where any mechanical energy generation equipment would produce noise that would give away its position.
Bismuth is already widely used in commercial thermoelectric semiconductors in the form of Bi2Te3 devices that boast a figure of merit, ZT, of around one. Higher figures of ZT reflect a generally higher thermodynamic efficiency.
Heavy elements like bismuth reduce a material's thermal conductivity, boosting its ZT, but using a material with a suitable bandgap is also important. Zide believes that dilute bismides will perform well on both fronts.
“My target ZT is an ambitious 2.5-3,” he said. “I think there is significant reason to think it's feasible, but it s one of those things we won t know until we get there.”
This is a departure from the little dilute bismide research performed to date, which has focused on adding bismuth to GaAs optoelectronic devices.
Dilute bismides allow bandgap narrowing like dilute nitrides, where the introduced nitrogen is more electronegative than arsenic and causes conduction band anticrossing. By contrast bismuth is electropositive, and causes bandgap narrowing via valence band anticrossing.
However, unlike dilute nitrides, GaAs1-xBix is not lattice matched to the GaAs substrate it is grown on, straining the material system and potentially making it bow and crack. Zide says attempts to reduce the strain by producing GaAsNBi compounds have actually made highly defective material, so instead he is looking to start from a different substrate.
Zide will also exploit an approach that he has developed to epitaxially incorporate metallic nanoparticles into InGaAs. These nanoparticles provide another way to reduce the semiconductor material s thermal conductivity.
