Future Materials


The thermoelectric materials research community consists mainly of numerous groups in universities and institutes in North America, Europe, the former Soviet Union, China and Japan. In addition, a handful of companies, including Marlow, contribute to the advancement of thermoelectric materials.

This community exchanges results and ideas at regular meetings of the International Thermoelectrics Society (ITS), the European Thermoelectrics Society, and the Thermoelectrics Society of Japan. Of these, the most heavily attended meeting is the annual International Conference on Thermoelectrics, which rotates from North America to Europe to Asia. The 27th ICT was held August 2-7, 2008, in Corvallis, Oregon, with approximately 360 participants.

Highlights from the annual ICT meeting provide a good snapshot of the state of the next generation of thermoelectric materials. A snapshot includes the following notable points: 

Nanostructured bulk materials

  • An MIT-Boston College team showed that the ZT of p-type bismuth telluride alloys can be improved above room temperature by making the material with grain size of ~ 20 nm. They believe preferential scattering of electrons, which are detrimental minority carriers, causes the improvement.
  • M. Kanatzidis of Northwestern University and co-workers continue to explore a handful of nanostructured materials based on PbTe, achieving good ZT values in both n-type and p-type. The materials generally are greater than 90% lead telluride, and nanostructuring occurs naturally after adding compounds such as AgSbTe2 or PbSe. The ZT improves, mainly due to reduced thermal conductivity, reaching to around 1.7 at 700 K in the best compositions.
  • Zhejiang University in Hangzhou China has a strong program on nanocomposite thermoelectric materials. They presented results on skutterudite alloys, bismuth telluride alloys and Mg2(Si,Sn). The skutterudite and bismuth telluride alloys were blended powders, with typical grain size of 10 nm prior to consolidation and 100 nm after consolidation. A peak ZT close to 1.5 was found for the nanocomposite Bi2Te3-Sb2Te3, which would be a significant improvement if confirmed.
  • Tritt et al. of Clemson University reported on consolidated PbSe-coated PbTe nanoparticles made by CVD. They found significant ZT enhancements, mainly due to reduced thermal conductivity. The ZT values of their control samples (no nanostructuring) were low enough that the enhanced ZT values were still below 1.


Thin film materials

  • Micropelt of Fraunhofer, Germany, makes thermoelectric devices based on sputtered films of Bi2Te3 alloys. At ICT 2008, they showed results for 18 and 36 micron thick films. For the thinner films, watt densities reach 70 W/cm2; for the thicker films, they achieve temperature differences of 60ºC, with the hot side at 85ºC.
  • Research Triangle Institute (RTI)and spin-off Nextreme make devices based on MOCVD films of Bi2Te3 alloys. They measure n-type and p-type average material ZT > 1 at room temperature, but film thickness is limited to less than 10 microns, and device ZT is well under 1 due to parasitic losses. RTI presentations focused on using their superlattice films in the coolest stage of cascaded power generation devices.
  • Cahill and Johnson


New compositions

  • A team led by J. Heremanns of Ohio State University reported that they enhanced ZT in PbTe by incorporating thallium. Experiments showed that the enhancement was due to an alteration of the density of states in the valence band, and theory suggests a resonance interaction between thallium and the energy levels of the PbTe host causes this alteration.
  • Skutterudite alloys based on CoSb3 continue to move forward and present themselves as an alternative to PbTe alloys for power generation applications with a hot side below 600ºC. The best ZT values now are around 1.2 and 1.0 for n-type and p-type respectively at 700 K.
  • A number of groups are studying Mg2(Si,Sn) alloys because n-type samples can have a good ZT (>1 at 650 K), and the constituents are very inexpensive and have low toxicity. Researchers report making good quality samples both by powder methods and by melt-growth methods.
  • Tl-(Ag or Bi)-Te ternary systems contain compounds that have high ZT values due to extremely low thermal conductivity values. A group from Japan presented a ZT = 1.2 at 500 K for Tl9BiTe6.
  • Semiconducting Half-Heusler compounds have formulas such as (Ti,Hf,Zr) NiSn and TiCoSb. These materials offer greater thermal and chemical stability than PbTe or skutterudite alloys, but their ZT values lag, especially for p-type. Y. Kimura of Tokyo Institute. of Technology reported ZT values around 0.9 at 600-700 K for n-type (Hf,Zr) NiSn grown by optical float zone.


Development of current materials (not nano-structured)

  • Two groups discussed extrusion of traditional Bi2Te3 alloys and reported ZT values close to those of more common melt-grown versions of these alloys.
  • N-type PbTe is a material that performs well and has acceptable mechanical properties. P-type PbTe-based materials have either disappointing ZT values (when Sn partially replaces Pb) or poor mechanical properties (when doped with Na, for example). Y. Gelbstein of Ben-Gurion University showed that Na-doped PbTe is almost perfectly brittle and argued that PbTe-SnTe alloys are therefore a more practical choice.
  • SiGe alloys were used in the hottest stage (450 to 850ºC) of a 3-stage cascade and contributed 5% to a total 3-stage efficiency of ~ 18%, according to Research Triangle Institute. These SiGe alloys were made by Ames Laboratory.
  • “TAGS” is a scrambled acronym for GeTe/AgSbTe2 alloys. The 85% GeTe composition has a high ZT and has been used for p-type legs in power generation systems for deep-space probes. With the discovery that AgSbTe2 forms nano-precipitates in PbTe, a few groups have revisited the structure of TAGS alloys. Some groups report no nanostructures, while others presented evidence of nanostructures. One group showed that the distribution of micro-precipitates in TAGS alloys was strongly dependent on thermal history.


Applicability of materials

  • H. Bottner of the Fraunhofer IPM (Freiburg, Germany) and J. Sharp of Marlow presented order-of-magnitude analyses to argue that tellurium is both too expensive and too rare to be useful for large-scale applications such as waste heat recovery. With tellurium-based materials, meeting cost targets might be impossible, and a single major successful application will quickly lead to depletion of the existing tellurium reserves. Similarly, elements such as Ge, In, and Ga cannot be major constituents in a power generation material intended for large-scale use.
  • Thallium is an extremely toxic element, so its presence in thermoelectric materials likely will limit their usefulness.
  • Thin film materials are a poor match for nearly all power generation applications due to a severe mismatch in watt density values.
  • Thin film materials might be useful for spot cooling of microelectronics or other very high watt density applications that could emerge, but the low ZT of thin film devices will limit their usefulness elsewhere.