Toward Waste Heat Recovery Using Nanostructured Thermoelectrics

Transcription

1 Toward Waste Heat Recovery Using Nanostructured Thermoelectrics Sanjiv Sinha Mechanical Science & Engineering University of Illinois at Urbana-Champaign

2 Potential for Waste Heat Harvesting University of Illinois November [1] Solar 0.11 Nuclear 8.35 Hydro 2.68 Wind 0.70 Geothermal 0.37 Exergy Cost > 11 Quads Rejected Energy Natural Gas Coal Energy Services Biomass 3.88 Petroleum ~ 57% of energy consumed in the US rejected as heat

3 Sources of Low Quality Waste Heat T exhaust ~ C T IN ~ 27 C (ASHRAE) dt ~ C (HD, State of art at Intel) T~ C Source: BMW T water ~ 33 C T flue 100 C (h furnace ~ 90%) Source: SMU h CARNOT (T H =100C) ~ 20% Difficult to extract work Fluid machines infeasible Transcritical CO 2, organic Rankine, Kalina are expensive Thermoelectrics? [2]

4 [3] Thermoelectrics Thermoelectric Generators Thomas Johann Seebeck ( ) e Seebeck, T.J., Magnetische polarisation der metalle und erze durch temperaturdifferenz, Verlag von Willhelm Engelmann, Leipzig, Germany (1895) S ~ k V /K N P h+ - + p-leg p-n Thermoelectric Junction Z 2 S k ~ ~ Thermopile (Nobili & Melloni) ~ 200 W maximum power Patent (Yamamoto) 5% efficiency, ZT (Abraham Ioffe) kw (Russia) Bi-Sb-Te-Se alloys Radioscope TEG (Pioneer, Voyager, Galileo, Cassini )

5 [4] BSST Thermoelectrics Applications Refrigeration Amerigon Power Generation Waste Heat Recovery Cassini BSST

6 The Thermoelectric Heat Engine Fraction of Carnot HOT Q in I T Hot T k -1 V=S T R Engine Work COLD T Cold h = 1- T cold T hot æ ç h = h CARNOT S ç è ZT = ç k S Q out ö 1+ ZT ZT + T C T H ø st k 2 Bi 2 Te [5] ZT

7 [6] The Z Conundrum Z = S2 s k Microscopic linkage between S, and k k ( k Electron + k Lattice ) Degenerate semiconductors S S 2 Carrier concentration Metals k L Carrier concentration k s = LT Wiedemann Franz Law k E k S

8 Conversion Efficiency [7] æ ç h = h CARNOT ç ç è 1+ ZT ZT + T C T H ö ø Bi 2 Te 3 alloys have ZT ~ 1 at 300 K n-type p-type Scalability Issues: Low abundance ~15x expensive as Si What about silicon? k ~ 150 W/mK (300 K) Bulk ZT ~ 0.01

9 Thermal Conductivity Lattice vibrations conduct heat in Si T h T c T h T c Phonons (Quantum) From Kinetic theory, k ~ C v L L limited by different scattering processes Boundary Scattering Impurity Scattering Phonon-Phonon Scattering (Umklapp) [8]

10 Silicon Thermal Conductivity [9] L PHONON ~ nm Boundary scattering dominates in 100 nm Si Impurity Umklapp Phonon boundary scattering is frequency dependent and tunable through roughness Boundary k B (300)

11 UIUC Silicon TE Device Concept [10] Hot Gas Cooling Water Sq. mm array > 25 % areal coverage Resistivity ~ mw-cm L > 10 mm Rms roughness > 2nm Corr. Length < 100 nm Dia < 100 nm

12 Wire Array Fabrication [11] MAC-Etch Image Reversal via Lift-Off

13 Device Formation [12] Controlled Roughness Enhancement Atomic 5nm 10nm 20nm SOG/Poly Fill & Contact Formation Transfer Printing (n- and p-type)

14 HRTEM Roughness Characterization [13] Roughened wire still crystalline Image Stitching Process

15 Electrical Conductivity Measurements [14] r can be reduced to ~1 mwcm via controlled doping

16 Single Wire Thermal Measurements [15]

17 Thermal Conductivity of Wire Arrays [16]

18 T/P (K/W) Array Seebeck and Thermal Measurements V/P (uv/w) Voltage pad Heater Substrate NW array q R ox R NW Sub f(w) V n V s q T h2 R ox Sub f(w) V s T = 0.34 K Frequency (Hz) Ref NW NW sample Frequency (Hz) Reference Ref. NW array V = 60 μv [17]

19 Summary and Outlook Silicon thermoelectrics potentially enables unprecedented cost-effective waste heat recovery Ongoing work at UIUC seeks to fabricate the first truly nanostructured silicon thermoelectric device Preliminary characterization work shows promising trend and focus is now on device engineering Cahill Fang Ferreira Li Rogers Sinha Keng Bruno Marc Jun Seok Numair Dongwook Myunghoon Krishna Jyothi Joe Feser Jae Cheol Jae Ha Chen Jun-Hwan University of Illinois November [18]