Nanoelectronic Thermoelectric Energy Generation Lourdes Ferre Llin l.ferre-llin.1@research.gla.ac.uk 1
Overview: Brief introduction on Thermoelectric generators. Goal of the project. Fabrication and Measurements for different devices. Thermal conductivity. Electrical conductivity. Seebeck coefficient. Conclusions and future work. 2
Thermoelectric devices: Are used in applications for generating electricity due to a difference of temperature and also for producing cooling in presence of electricity. There are three stablished thermoelectric effects, known as: Peltier effect. Thomson effect. Seebeck effect. Heat source metal contact The Seebeck effect is the responsible for power generation. Their fabrication consists of pairs of p-type and n-type semiconductor materials forming a thermocouple. n-type metal p-type metal Heat sink I Load 3
How does it work? These thermocouples are then connected electrically forming an array of multiple thermocouples. When we apply heat and cold, it is generated a difference of temperature that will generate electricity. 4
Thermoelectric generators. Applications: Almost any heat source could be used to generate electricity, such as natural sources (solar heat) and waste heat of any device or machine that generates heat as a byproduct. Recovering the energy lost as heat could improve drastically the efficiency of a device or machine. Problem of thermoelectric devices to date: Low efficiency. Commercial devices produced using Bi and Te. Advantages: Not moving parts. Require very little maintenance Can provide energy as long as there is a difference of temperature. 5
Objective of this project. Fabricate micro/nano structures to study the improvement of the efficiency. The heterostructures technology made of Si/SiGe that will be investigated are: 2D superlattices. 0D quantum dots. 1D nanowires. The final thermoelectric design will be integrated on a mm-sized single silicon chip. This will be used to power a CMOS sensor. The generator will work as a power source for an autonomous system 6
Efficiency & Figure-of-Merit The efficiency of a generator is given by: θ = energy derived to the load heat energy absorved at hot junction Maximum efficiency θ max = η c λ λ = η c = T H T C T H 1+ Z T 1 1+ Z T + T C Product of the Carnot efficiency times the thermoelectric properties of the materials. Figure-of-Merit Z = T H α 2 σ K 7
α2 σ Z= K 2D Superlattice Measure thickness of the Multi-QW layer each time we process a new sample. This thickness changes from 5 um to 10 um. August 5th, 2011 Multi-QW structure Silicon substrate NiPS Workshop 2011 Thin Si SOI wafer 8
Thermal conductivity (W m 1 K 1 ) 3-omega method: Placing a directly conductor on the surface of the material, which will serve as a heater and as a thermometer. Driving an AC current through this line at frequency 1-omega. This current will heat the conductor, which in return will be measurable as a resistance change at the frequency 3-omega. ΔT = 2 dt dr R V 1w V 3w Allows to measure the temperature oscillation of the line. 9
Thermal conductivity: 3-omega method - Anisotropic material: 1D heat model Vertical and horizontal contribution Heaters: 10 nm NiCr + 50 nm Au - To measure the V_3w we need to cancel the voltage at 1w AC 20 ohms + - Lock-in PC In phase Out-of-phase m -Different widths for the line, from 5um up to 200um. -Length of the line variable to change the resistance of the metal line. -Difficulty to measure the V_3w that is typically one thousandth off the primary voltage V_1w 10
Thermal conductivity: 3-omega method Analysis Slope-method for low frequencies: measuring K for substrate. Passivation layer Heater Thin-film Substrate 150 Silicon wafer. Line width 200 um In phase Out of phase 100 Out-of-phase constant (T ) = P! " K "(3 2 # $ # ln% 2 # i "! 4 ) Temperature Oscillation (K) 50 0 50 Thermal conductivity for silicon 143(W! m "1! K "1 ) 100 - Only valid for small arguments - Not dependant of the frequency 150 Frequency (Hz) 10 3 11
Thermal conductivity: 3-omega method Analysis Analysis Differential Differentialmethod method for forhigh highfrequencies: frequencies:! Superlattice!= K " # Cp q!1 = SOI " 2#w Penetration depth. Design 1 (p type) 30 Penetration depth ( um ) 25 20 10 um SL etched 15 Heat Source T1 Heat Source T2 10 5 0 10K August 5th,2011 2011 May 24th, Thursday, 4 August 2011 Tuesday, 24 May 2011 20K 30K Frequency ( Hz ) 40K 50K 60K NiPS Workshop 2011 Group Meeting 12 13
Thermal conductivity: 3-omega method 15.8 15.6 Silicon substrate Silicon plus QW Temperature Oscillation (K) 15.4 15.2 15 14.8 14.6 14.4 1.27 k f = P! d f 2! b!("t s+ f (#) $ "T s (#)) - Cross check results with other technique. 14.2 14 5 0.4 13.8 6K 7K 8K Frequency (Hz) Thermal AFM 13
Thermal conductivity: Thermal AFM Pd resistor Thermal probe fabricated in JWNC. Palladium resistor located at the end of the tip. Four gold lines that allow a standard four-point measurement. Set up: A fixed temperature is needed as a reference. Reference obtained by calibrating the system. NiCr Heater Instrument used for calibration is able to define an absolute temperature based on the measurement of Johnson noise. Au Shield NiCr resistor Dot with known temperature used to calibrate the thermal probe. 14
Thermal conductivity: Thermal AFM! Fabrication: " Passivation layer (15 nm) " Deposition of NiCr (33 nm) + Gold (100 nm) " Etching gold to open a window with only NiCr 500 um 301 300.5 Temperature Oscillation. Line 5 um. Line width 25 um Curve fitting. Line width 25 um Line width 5 um Curve fitting. Line width 5 um 33 nm NiCr Square under test 300 Temperature ( K ) 299.5 299 298.5 298 297.5 Bigger contribution by the lateral thermal conductivity K 6.36 297 0 2 4 6 8 10 12 14 16 18 20 22 Position ( um ) May August 24th, 5th, 2011 2011 NiPS Group Workshop Meeting 2011 16 15
Electrical conductivity Interested in measuring the lateral electrical conductivity. Fabrication: Development of an etch profile to contact all the QWs. Deposition of SiN as a passivation layer. Open windows on SiN to create ohmic contacts. Deposit metal. Multi QW layer Ohmic contact created by sputtering Aluminium SiN Al 16
Electrical conductivity Mixed process etch good profile for the side walls. Still the presence of an undercut at the top part of the side wall. Cut none metal Aluminium Aluminium SL cross section 17
Electrical conductivity Hall Bar devices!v 85 um I I+ I- Sample_ID Design p-type N A (cm 3 ) Thickness ρ(ω µm) σ(s m) 1 10 19 8557_J6 2 3.268 um 9.54 104,733 1 10 18 8569_I8 2 6.048 um 26.14 38,255 8482_E4 2-11.340 um 299.14 3,343 1 10 19 8572_D2 2 6.804 um 53.024 18,859 1 10 19 8579_E2 1 6.048 um 11.327 88,284 18
Seebeck coefficient/thermopower Voltage produce across two points on a material divided by the temperature difference between them. Standard Hall Bar device with the addition of having two heaters and two thermometers on top of the bars. This design allows to measure the three parameters on the same device. α = ΔV ΔT Pads to measure voltage Heater Thermometers 19
Seebeck coefficient/thermopower V V1-V2 20
Seebeck coefficient/thermopower V T1 T2 TCR 21
Seebeck coefficient/thermopower Strain balanced sub required to make Problems: actual measurements on the Thermal and Seebeck. Results show a big influence of the substrate thermal conductivity. Silicon 150 (W m 1 K 1 ) Necessary to remove the substrate of the final devices. SiGe Hall Bar with substrate removal Very thin membranes, easy to break them as well due to the stressed material under test. Strain balanced sub required to mak actual measurements on the Therm and Seebeck. August 5th, 2011 NiPS Workshop 2011 22
Conclusions and Future work Optimise substrate etch in order to get accurate results. Cross check thermal conductivity results with the thermal AFM, to know the error of our measurements. Design and fabricate future devices for measuring the vertical thermal and electrical properties of the different wafers obtained so far. Easier fabrication process as it will not be necessary to remove the substrate. 23
Participants in the project University of Glasgow Device processing Material characterisation Politecnico di Milano, Como Italy Epitaxial growth (LEPECVD) University of Linz, Austria Material characterisation (XRD) ETH in Zurich, Switzerland Material characterisation (TEM) 24
Thanks! 25