1. Thermoelectric Energy Transport in Nanostructures Ali Shakouri Baskin School of Engineering, University of California, Santa Cruz Http://quantum.soe.ucsc.edu Int. Workshop on Nanoscale Energy Conversion and Information Processing Devices Nice, France, 24 September 2006 Acknowledgement Postdocs/Students: Zhixi Bian, James Christofferson, Mona Zebarjadi, Rajeev Singh, Xi Wang, Daryoosh Vashaee, Yan Zhang, Kazuhiko Fukutani, Tammy Humphrey Collaborators: John Bowers, Art Gossard, Arun Majumdar, Venky Narayanamurti, Rajeev Ram, Tim Sands, Avi Bar-Cohen, Stefan Dilhaire, Ed Croke, Peidong Yang, Holger Schmidt Sponsors: ONR/MURI, Intel, Canon, National, Packard Foundation, DARPA/Heretic, NSF

2. 2 AS 9/24/2006 Motivation: Microprocessor Evolution Electronic/Optoelectronic devices → Generate high/ localized heat density

3. 3 AS 9/24/2006 Possible Applications Waste heat recovery Electric power generator with no moving part Microscale power sources Direct Conversion of Heat into Electricity Significant amount of heat generated as by product of any energy conversion. Thermal Electrical Optical Magnetic Mechanical Chemical/biological

4. 4 AS 9/24/2006 Rejected Energy 61% Total 91.4 quad (↑ x3) 1950

5. 5 AS 9/24/2006 a b a I Q Q Peltier: Peltier and Seebeck Effects Thomson: Commercial TE Module  T=72C Cooling density <10W/cm 2 Efficiency 6-8% of Carnot RTGs (space power) Seebeck: a b V T1T1 T2T2 a

6. 6 AS 9/24/2006 Efficiency of TE Power Generation ZT = 1.2-3.6 ZT = 0.3-0.9 Efficiency (COP) depends on a single ratio ( Z )

7. 7 AS 9/24/2006 PbTe/PbTeSe Quantum Dot Superlattices Ternary: ZT=1.3-1.6 Quaternary: ZT=2  T=43.7 K, Bulk  T=30.8 K T.C. Harman, Science, 2002  T=32.2 K, ZT ~2-2.4 R. Venkatasubramanian, Nature, 2001 Nanostructure Bulk Power Factor (  W/cmK 2 ) 25.5 28 40 50.9 Thermal Conductivity (W/mK) 0.5 2.0 0.5 1.26 PbTe/PbSeTe Bi 2 Te 3 /Sb 2 Te 3 Superlattice Bulk In-plane geometry Cross-plane geometry (From M. S. Dresselhaus, Rohsenow Symposium, 2003) Superlattices/ Quantum Dot Thermoelectrics T. C. Harman (2002) and R. Venkatasubramanian (2001)

8. 8 AS 9/24/2006 Thermionic Emission for Energy Conversion Cathode Barrier Anode Energy Hot electron Cold electron Metal/Semiconductor Superlattice, Embedded Nanostructures Low work function Vacuum (ions) Low work function Metal/ Deg. Semicond Solid-State Vacuum Hot Cold Selective emission of hot electrons over a potential barrier can generate electrical power from temperature difference Thermodynamic reverse process: evaporative cooling of electrons

9. 9 AS 9/24/2006 Si/SiGeC Superlattice Structures for Heterostructure Thermionic Filtering 150x SiGeC/Si Superlattice (10nm/10nm) Barrier Si Cathode Si (001) Substrate Anode MBE Grown 5” Substrate Material and Processing Compatible with SiGe HBTs. 1 µm Hot Electron Cold Electron Funded by ONR and DARPA/ARMY HERETIC Si Si 0.89 Ge 0.1 C 0.01 X. Fan, E.Croke, J.E. Bowers, A. Shakouri, et al., “SiGeC/Si superlattice micro cooler,” Applied Physics Lett. 78 (11), 2001. Featured in Nature Science Update, Physics Today, AIP April 2001

10. 10 AS 9/24/2006 Microrefrigerator on a chip Temperature resolution: 0.006 o C Spatial resolution: submicron High resolution thermal imaging Thermal imaging camera; J. Christofferson, A. Shakouri, Review of Scientific Instruments Feb 2005. Nanoscale heat transport and microrefrigerators on a chip; A. Shakouri, Proceedings of IEEE, 2006 Maximum cooling: 4C (300K), 12 (500K) Cooling power density: >500 W/cm 2 Response time: < 20-40  s Materials: SiGe, SiGeC, InGaAs, InP Fabrication: IC compatible ZT~0.08-0.1

11. 11 AS 9/24/2006 J. Snyder (2003) http://www.its.caltech.edu/~jsnyder/thermoelectrics/science_page.htm I For almost all materials, if doping is increased, electrical conductivity increases but Seebeck coefficient is reduced. S S2S2 How to improve ZT?

12. 12 AS 9/24/2006 Energy Density of States E f High doping Doped Bulk Semiconductor/ Metal Highly-Doped Tall Barrier Superlattice EfEf E barrier Metallic Superlattices for Thermionic Energy Conversion Distance Energy Symmetry of DOS near Fermi energy is the main factor determining Seebeck coefficient. E f Low doping

13. 13 AS 9/24/2006 Program Manager: Mihal Gross D. Vashaee., A. Shakouri, Physical Review Letters March 12, 2004 Non-planar Barrier UCSC Berkeley Harvard MIT NCSU Purdue UCSB Director: A. Shakouri ZT for metallic superlattices with non-planar barriers Thermionic Energy Conversion Center MURI Assume:  lattice =1W/mK, mobility ~10 cm 2 /Vs Planar Barrier Planar barriers are not ideal for hot electron filtering. ZT>5 is possible with metallic structures with non-planar barriers. Hot and cold electrons in equilibrium Hot electron filter

14. 14 AS 9/24/2006 TEM/HAADF of Semimetallic ErAs Nanoparticles in InGaAs Matrix In,Ga As Er STEM images show that the ErAs particles have the rock salt structure. The As sublattice is continuous across the interface. STEM images show that the ErAs particles have the rock salt structure. The As sublattice is continuous across the interface. 110 001 1nm D. O. Klenov, D. C. Driscoll, A. C. Gossard, S. Stemmer, Appl. Phys. Lett. 86, 111912 (2005) HAADF

15. 15 AS 9/24/2006 0.4 ML 40 nm 0.1 ML 10 nm In 0.53 Ga 0.47 As Kim et al., Physical Review Letters, 30, 045901 (2006) In 0.53 Ga 0.47 As 0.3 % ErAs 3.0 % ErAs 3.0 % ErAs:In 0.53 Ga 0.28 Al 0.19 As Thermal Conductivity of ErAs:In 0.53 Ga 0.47 As

16. 16 AS 9/24/2006 ErAs: InGaAs/InGaAlAs SL n-InP substrate 50nm 5E18 n-InGaAs 20nm n-InGaAs/ErAs 0.3% 10nm InGaAlAs 20nm n-InGaAs Cap layer Sample 1 –1E19 Sample 2 –4E18 Sample 3 –2E18 Add superlattice energy filtering to increase the thermoelectric power factor. Joshua Zide, Daryoosh Vashaee, Gehong Zeng et al., submitted to PRB 2006 70x

17. 17 AS 9/24/2006 Cross-plane/ In-plane Seebeck Characterization J. Zide et al., (UCSB, UCSC) submitted to Physical Review B, 2006 Theory/Experiment Seebeck II, ┴ (300K) ErAs: InGaAs/InGaAlAs Superlattices 1e19 cm -3 2e18 cm -3 6e18 cm -3 Theoretical ZT ZT Temperature (K)

18. 18 AS 9/24/2006 Thermoelectric single element characterization ErAs generates significantly more power despite the lower effective Seebeck BiTe degrades rapidly at higher temperatures while ErAs improves with temperature 050100150200250300350 0 1 2 3 4 5  T (K) Power density (W/cm 2 ) -ErAs (SL+substrate) -SiGe (SL+substrate) -BiTe (bulk) 160x(10nm (InGaAs) 0.6 (InAlAs) 0.4 /20nm (n-InGaAs) 0.97 Er 0.03 ) on 474  m doped InP substrate 200x(75Å SiGe 0.16 / 75Å SiGe 0.24 ) on 403  m doped Si substrate T hot V+V+ V-V- Heat er OFHC copper TE sample Al cold plate Chilled water Ceramic rails (insulating) Peter Meyer, Rajeev Ram (MIT)

19. 19 AS 9/24/2006 Thin film array generator 200 n-p couples, 5-10 microns ErAs:InGaAs/InAlAs superlattice thin films, 120x120  m 2, 12 ohm load G. Zeng, J. Bowers, et al. (UCSB) Appl. Physics Letters 2006

20. 20 AS 9/24/2006 Monte Carlo + Poisson Equation InGaAs InGaAsP InGaAs Heat Sink Anode Bias Hot Source Cathode Cathode contact layer Anode contact layer Barrier (main- layer) + Electron-phonon energy exchange (S) Goal: Range of validity for thermoelectric and thermionic transport formalisms Mona Zebarjadi, Keivan Esfarjani, Ali Shakouri (UCSC)

21. 21 AS 9/24/2006 Electron-phonon energy exchange (S) Peltier CoolingPeltier Heating Non-equilibrium transport in the barrier Energy relaxation length in cathode Energy relaxation length in anode Mona Zebarjadi, Keivan Esfarjani, Ali Shakouri (UCSC)

22. 22 AS 9/24/2006 Effective Seebeck Coefficient vs. Barrier Thickness Convectional Thermoelectric transport Convectional Thermionic transport Mona Zebarjadi, Keivan Esfarjani, Ali Shakouri (UCSC)

23. 23 AS 9/24/2006 Summary Micro-refrigerator on a chip Cooling 4 -7C, >500W/cm 2, 20-40  s Solid-state thermionic energy conv. Metallic SL and embedded nanoparticles ErAs: lattice thermal conductivity 6→ 2-3 W/mK Increase ┴ Seebeck coefficient 200→600  V/K Power generation 1 element >5W/cm 2 for  T=300C Improvement in ZT: decouple S, , k Microscopic origin of TE/TI –Location and spatial extent of regions where Peltier cooling/ heating occurs –Transition from TE to TI transport Statistical properties of reservoirs Students/postdocs UCSC Zhixi Bian, Rajeev Singh, Mona Zebarjadi, Yan Zhang Younes Ezzahri, Daryoosh Vashaee, Tammy Humphrey Berkeley Woochul Kim Susanne Singer Harvard Kasey Russel MIT Peter Mayer Purdue Vijay Rawat UCSB Josh Zide, Gehong Zeng, J-H Bahk Acknowledgement: ONR MURI (Dr. Mihal Gross), Packard, DARPA, Intel, Canon SUMMARY

24. 24 AS 9/24/2006 k B T h ~ 75 meVk B T c ~ 25 meV Why there is Carnot limit? T=900K T=300K Average Random Kinetic Energy of Carriers If an electron is moved from hot reservoir to cold reservoir with “no dissipation”, on the average the maximum amount of energy per electron available to do work is: (K B T h -K B T c )/K B T h = (T h -T c )/T h Carnot limit Ali Shakouri, TE, TI and TPV energy conversion, MRS Fall 2005

25. 25 AS 9/24/2006 k B T h ~ 75 meVk B T c ~ 25 meV h BlackBody ~ 400 meV Thermoelectric/Thermionic vs. TPV n p h BlackBody ~ 125 meV T=900K T=300K Photons emitted from hot source have higher average energy than electrons emitted at the same temperature.

26. 26 AS 9/24/2006 T=625C T=25C Photon-Assisted Thermionic Power Generation Solid State TI Possibility to use both hot electrons and hot photons? Ali Shakouri, TE, TI and TPV energy conversion, MRS Fall 2005

27. 27 AS 9/24/2006 Possibility to use phase transition (change in internal degrees of freedom, latent heat) in electron gas to improve TE energy conversion efficiency? Is there room temperature phase change for electrons?