Measurement of Seebeck coefficient perpendicular to SiGe superlattice

Transcription

1 Measurement of Seebeck coefficient perpendicular to SiGe superlattice Yan Zhang 1, Gehang Zeng 2, Rajeev Singh 1,James Christofferson 1, Edward Croke 3, John E. Bowers 2 and Ali Shakouri 1, 1 Electrical Engineering Dept. of University of California, Santa Cruz, CA 94056, 2 Electrical and Computer Engineering Dept. of University of California, Santa Barbara, CA 93106; 3 HRL laboratories, LLC Malibu, California Abstract Seebeck coefficient is one of e key parameters to evaluate e performance of ermoelectric coolers. However, it is very difficult to directly measure Seebeck coefficient perpendicular to in film devices because of e difficulty of creating a temperature gradient and measuring localized temperature and voltage change simultaneously. In is paper, a novel meod is described and it is used to measure e Seebeck coefficient of SiGe superlattice material perpendicular to e layers 1. Successful measurement was achieved by integrating a in film metal wire as a temperature sensor and heat source on top of e SiGe superlattice micro coolers. Extensive ermoreflectance imaging characterization was performed to ensure uniform temperature distribution on top of e in film device. Details of e experimental set-up and measurement technique are discussed. By analyzing e measured ermoelectric voltage for various device sizes and superlattice ickness, Seebeck coefficient of e superlattice material perpendicular to e layers is deduced. Introduction In VLSI circuits, high heating power density is one of e bottlenecks at limits e reliability and performance of e chip in high-speed, high-density applications. Conventional bulk Bi2Te3 coolers 2 have limited applications in microelectronic circuits due to low cooling power density and difficulty of integration and packaging 3. Conventional Si or III-V based semiconductor materials have a low ermoelectric figure -of-merit and ey are not suited for cooling application. There have been several excellent recent studies on BiTe-based in film coolers wi high ermoelectric figure-of-merit 2. We are concentrating on SiGe-based coolers for e possibility of direct integration wi silicon circuits. Use of ermionic emission in heterostructure superlattices can improve e figure-of-merit by selective emission of hot electrons above potential barriers and by reducing e phonon heat conduction in multiplayer materials. SiGe and SiGeC-based coolers have already demonstrated a cooling power density exceeding 500W/cm 2 and a maximum cooling of 4 0 C at room temperature. 4 In is paper, we focus our study on e Seebeck coefficient perpendicular to SiGe superlattice layers. Alough ere are some papers on e cross-plan ermal conductivity, electrical conductivity of SiGe superlattice 5,6,7,8. There are still very few papers on e experimental measurements of e Seebeck coefficient perpendicular to e in film because of its difficulty to measure e voltage and temperature change simultaneously. 9,10 We used an integrated in film resistor bo as a heater and a sensor on top of e SiGe superlattice microcooler. The processing was done using e standard semiconductor fabrication process 11. In a linear transport regime, e Seebeck coefficient is defined as e voltage produced across two points on a material divided by e temperature difference between em. This has an expression of S = 12. Seebeck T coefficient is e main parameter to evaluate performance of ermocouples. It is also e key parameter to calculate e ZT, e ermoelectric figure of merit, which has e express ion of: 2 ZT = S T / ρk T (Eqn.1) Where S: Seebeck coefficient; T: temperature; ρ: electrical resistivity; K T: Thermal conductivity 12. Experiments The micro-cooler structure is based on cross-plane electrical transport eory. The main part of e cooler is a superlattice structure of 80Å Si/40 Å Si 0.7Ge0.3 grown at C, doped wi boron to about 5x10 19 cm -3. The buffer layer was grown on top of e silicon wi e structure of 1um SiGe 0.1 doped to 5x10 19 cm -3 and 1µm SiGe 0.1 /SiGe 0.15 C Finally, e sample was capped wi 250nm SiGe 0.1, doped to approximately 2x10 20 cm -3. The SiGe/Si micro-coolers are fabricated wi standard silicon integrated circuit technology. The cooler device areas were defined by etching mesas down to e SiGe buffer layer. Ti/Al/Ti/Au metallisation was made on top of e mesa and on e SiGe buffer layer next to e mesa for top and bottom contacts respectively. The size of e sample is ranging fro m 50x50µm to 100x100µm. Figure. 1 shows a scanning electron micrograph picture of is device. Page 1

2 Report Documentation Page Form Approved OMB No Public reporting burden for e collection of information is estimated to average 1 hour per response, including e time for reviewing instructions, searching existing data sources, gaering and maintaining e data needed, and completing and reviewing e collection of information. Send comments regarding is burden estimate or any oer aspect of is collection of information, including suggestions for reducing is burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA Respondents should be aware at notwistanding any oer provision of law, no person shall be subject to a penalty for failing to comply wi a collection of information if it does not display a currently valid OMB control number. 1. REPORT DATE REPORT TYPE 3. DATES COVERED to TITLE AND SUBTITLE Measurement of Seeback coefficient perpendicular to SiGe superlattice 5a. CONTRACT NUMBER 5b. GRANT NUMBER 5c. PROGRAM ELEMENT NUMBER 6. AUTHOR(S) 5d. PROJECT NUMBER 5e. TASK NUMBER 5f. WORK UNIT NUMBER 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Department of Electrical and Computer Engineering,University of California,Santa Barbara,CA, PERFORMING ORGANIZATION REPORT NUMBER 9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR S ACRONYM(S) 12. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release; distribution unlimited 13. SUPPLEMENTARY NOTES The original document contains color images. 14. ABSTRACT 15. SUBJECT TERMS 11. SPONSOR/MONITOR S REPORT NUMBER(S) 16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT a. REPORT b. ABSTRACT c. THIS PAGE 18. NUMBER OF PAGES 4 19a. NAME OF RESPONSIBLE PERSON Standard Form 298 (Rev. 8-98) Prescribed by ANSI Std Z39-18

3 calibrated by measuring e resistance wi very low excitation currents at different ambient temperatures. Variations of e resistance wi temperature could be fitted well wi a line, wi an error less an 0.02%. The heater samples used in is experiments have an average of 4x10-3 dr/ / 0 C in a unit resistance ( dt R ). Subsequently ermoreflectance imaging was used to measure e temperature distribution on top of e device when e heater was on. As it can be seen from Fig. 3, e heating was localized on top of e in film device. Figure 1 The SEM picture of SiGe superlattice microcooler integrated wi heater sensor First resistance versus temperature of e heater sensor was calibrated. Four -wire measurement was used to reduce e effect of contact wires and pads. Then at a given heater power (fixed top layer temperature, T h ), e voltage difference across e device was measured. The bottom of e silicon substrate was maintai ned at e heat sink temperature (T s ). It is important to note at e measured ermoelectric voltage has a contribution from bo e superlattice Seebeck coefficient and also e Si substrate Seebeck coefficient. (Eqn.2) V = S1 ( Ts ' Ts ) + S2 ( Th Ts ') S1 is e Seebeck coefficient of bulk Silicon; Ts is e temperature at e interface between e superlattice layer and substrate. If e temperature T s is equal to T s, en e effective Seebeck coefficient equals to at of superlattice. However, is is not e case for our devices. We will see at e superlattice Seebeck coefficient S2 could be easily derived by analyzing experimental results for different in film layer icknesses and device sizes. Bao Yang et al.have used AC meod to measure Seebeck coefficient of in film s 13, e signal of our samples are large enough for an accurate DC measurements. Figure 3 Demonstration of localized heating of SiGe superlattice micro coolers Figure 4 displays e voltage change across e sample as a function of e temperature gradient for all different size samples, ranging from 50x50µm to 100x100µm. The slope of e curve is e average Seebeck coefficient wi contribution from superlattice and silicon substrate. It is know n at seebeck coefficient is geometry independent erm odynamic property. This corresponds wi e results in Figure 4, which shows size independent of effective seebeck coefficient. Table 1 summarizes e ermoelectric voltage measured in e experiments. It shows e sample SiGe 0.2 :B wi 3um superlattice ickness and doping concentration of 5-7x10 19 cm -3 has an average Seebeck coefficient of 135.4µV/ 0 C. Figure 2 Cross-section schematic of superlattice device (not to scale) Results and Discussion The heater resistance changes linearly wi temperature near 300K, so it was used as a temperature sensor on top of e superlattice device. The temperature sensor was Page 2

4 Delta V (mv) x50mm 2 60x60mm 2 70x70mm 2 100x100mm 2 R * 1 d = β A (Eqn.3) it is obvious to see R (SL): ermal resistance superlattice is proportional to e uperlattice ickness d sl. The change of R (SL) directly influences T s : R ( Si) ' R ( Si) R ( SL) Ts = Ts + * T + And it causes e change of effective seebeck coefficient: S= [S2*(Th-Ts )+S1*(Ts -Ts)]/DT (Eqn.5) (Eqn. 4) Temperature Difference (K) Figure 4 Thermoelectric voltage versus temperature difference between top of e device and bottom contact Delta V (mv) mm 3mm 6mm Sample size (µmxµm) Effective Seebeck coefficient (µv/ 0 C) 50x Temperature Difference (K) Figure 5 Comparison of seebeck coefficient wi different superlattice ickness (fitted) 60x x x Average Table 1 Si0.7Ge0.3 superlattice microcooler Seebeck coefficient data for different size samples Venkatasubramanian s recent paper in Nature 2 reported e cooling performance is related to e ickness of e superlattice layer. To furer investigate e influence of different superlattice ickness on e device property. We also measured e ermoelectric voltage for different superlattice ickness, 1um, 3um, and 6um. The results were illustrated in. It can be seen at icker e superlattice ickness, higher e measured ermoelectric voltage 135µV/ 0 C, 154µV/ 0 C and 174µV/ 0 C for 1µm, 3µm and 6µm superlattice respectively. The increase of e ermoelectric voltage is due to e increase in e ermal resistance of e in film device wi respect to e substrate as superlattice becomes icker. As illustrated in Figure 6 simplified ermal model of superlattice micro-cooler device, from e equation of Figure 6 Simplified ermal model of superlattice microcooler (This model assumes e ermal resistance of e SiN x and buffer layer is much smaller an e ermal resistance of e superlattice layer) To calculate e actual Seebeck coefficient of superlattice, we need to know e Ts, which is related wi e ermal resistance of e device. In is calculation, ermal resistance was obtained experimentally by using equation: T R = (Eqn. 6), T is e temperature difference created, Q Q is e heat load on top of e device. The ermal resistance for a 100x100um 2 device based on bulk silicon R (Si) was K/W, e measured ermal resistance of e device wi different superlattice ickness were listed in Table 2. From e known of ermal resistance of bulk Silicon and at of e overall device (superlattice+substrate), e T s - T s could be calculated by Eqn.4 assuming a temperature change of 10K. The of e device could be obtained by its effective Seebeck Page 3

5 cofficient. Then a pure superlattice of is structure S2 could be calculated by Eqn.2 wi all e known dat a. Table 2 lists all e calculation results and it showed an average seebeck coefficient of is structure superlattice is around µv/ 0 C. SL ickness R (Device) (K/W) (mv) T s -T s (K) coefficient µv/ 0 C Table 2 Calculation of superlattice ickness ( Sample size 100x100µm) S (SL) (µv/ 0 C) 1um um um Ave. SL seebeck 10 T. Yao, Appl. Phys. Lett. 51, 1798 (1987). 11 Xiaofeng Fan etc. Applied Physics Letters, Vol. 78, No.11, 12 March CRC Handbook of Thermoelectrics, ed. by D.M. Rowe (CRC Press, Boca Raton, FL, 1995) 13 Bao Yang et. al. Characterization of cross-plane ermoelectric properties of Si/Ge Superlattices Proceeding of 20 International Conference on Thermoelectrics (2001) Conclusions We showed at e integrated heater sensor on top of e in films provides a convenient meod to characterize e Seebeck of SiGe superlattice material. This will help furer development and improvement of e performance of SiGe superlattice coolers. Seebeck coefficient of ~217µV/K was measured for Si/SiGe superlattice p-doped to 5x10 19 cm -3. The experimental results were verified by obtaining consistent results wi various superlattice ickness (1-6um) and device sizes (2,500-10,000 µm 2 ). Acknowledgements This work was supported by e DARPA HERETIC program and e Army Research Office. Reference 1 Xiaofeng Fan etc. Applied Physics Letters, Vol. 78, No.11, 12 March Rama Venkatasubramanian et. al. Thin-film ermoelectric devices wi high room-termperature figures of merit, P597 Vo. 413, October 11, 2001, Nature 3 Xiaofeng Fan et. al. Integrated Cooling for Si-based Microelectronics, International conference on Thermoelectrics, June Xiaofeng Fan et. al. High cooling power density SiGe/Si micro coolers Elec. Lett., Vol. 37, No. 2, 18 January T. Borca-Tasciuc et.al. Superlattices and Microstructure 28, 199 (2000) 6 R. Venkatasubramanian, Phys. Rev. B 61, 3091 (2000) 7 H. Beyer, et al Proc. 18 Int. Conf. Thermoelectrics, ICT 99, 687 (1999) 8 R. Venkatasubramanian, Recent Trends in Thermoelectric Materials Research III, T. M. Tritt, Ed., Academic Press 71, 196 (2001) 9 W. L. Liu et. al. J. Nanosci. And Nanotech. 1, 37 (2001) Page 4