Thermal Stability Of Thermoelectric Materials Via In Situ Resistivity .
Thermal stability of thermoelectric materials via in situ resistivitymeasurementsK. C. Lukas, W. S. Liu, Q. Jie, Z. F. Ren, and C. P. OpeilCitation: Rev. Sci. Instrum. 83, 115114 (2012); doi: 10.1063/1.4767904View online: http://dx.doi.org/10.1063/1.4767904View Table of Contents: hed by the American Institute of Physics.Related ArticlesBX90: A new diamond anvil cell design for X-ray diffraction and optical measurementsRev. Sci. Instrum. 83, 125102 (2012)Experimental verification of the 3-step model of photoemission for energy spread and emittance measurementsof copper and CsBr-coated copper photocathodes suitable for free electron laser applicationsAppl. Phys. Lett. 101, 231103 (2012)Electrical ammeter based on spin-valve sensorRev. Sci. Instrum. 83, 105113 (2012)Note: Development of 9 A current source for precise resistance measurement methodRev. Sci. Instrum. 83, 106105 (2012)Four-probe electrical measurements with a liquid pressure medium in a diamond anvil cellRev. Sci. Instrum. 83, 103902 (2012)Additional information on Rev. Sci. Instrum.Journal Homepage: http://rsi.aip.orgJournal Information: http://rsi.aip.org/about/about the journalTop downloads: http://rsi.aip.org/features/most downloadedInformation for Authors: http://rsi.aip.org/authorsDownloaded 09 Dec 2012 to 136.167.96.29. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights and permissions
REVIEW OF SCIENTIFIC INSTRUMENTS 83, 115114 (2012)Thermal stability of thermoelectric materials via in situ resistivitymeasurementsK. C. Lukas, W. S. Liu, Q. Jie, Z. F. Ren, and C. P. OpeilDepartment of Physics, Boston College, Chestnut Hill, Massachusetts 02467, USA(Received 8 June 2012; accepted 2 November 2012; published online 28 November 2012)An experimental setup for determining the electrical resistivity of several types of thermoelectricmaterials over the temperature range 20 T 550 C is described in detail. One resistivity measurement during temperature cycling is performed and explained for Cu0.01 Bi2 Te2.7 Se0.3 and a secondmeasurement is made on Yb0.35 Co4 Sb12 as a function of time at 400 C. Both measurements confirmthat the materials are thermally stable for the temperature range and time period measured. Measurements made during temperature cycling show an irreversible decrease in the electrical resistivity ofCu0.01 Bi2 Te2.7 Se0.3 when the measuring temperature exceeds the maximum sample fabrication temperature. Several other possible uses of such a system include but are not limited to studying theeffects of annealing and/or oxidation as a function of both temperature and time. 2012 AmericanInstitute of Physics. hermoelectric materials have been widely studied overthe last two decades with most of the research focused on increasing the dimensionless figure-of-merit ZT, ZT S2 T/ρκ,where S is the Seebeck coefficient, ρ is the electrical resistivity, κ is the thermal conductivity, and T is the absolutetemperature.1, 2 However, ZT is not the only important parameter, especially when fabricating these materials for practical use, which is the ultimate goal. The materials should berelatively nontoxic, inexpensive, but most importantly theirphysical properties should remain stable over their temperature range of operation as well as during temperature cyclingwhich most materials will experience in waste heat applications for cars,3 solar panel use,4 etc.It is imperative when fabricating thermoelectric materialsto ensure the materials are thermally stable. Much time andeffort goes into the optimization of different synthesis parameters, such as pressing temperature, annealing temperature,annealing time. Thermoelectric samples measured in thiswork are prepared via ball milling and dc hot pressing techniques described previously.5 Studies discussing the transportproperties of thermoelectric materials often include thermalstability information based on temperature cycling, wherethe material is measured at different individual temperaturesa number of times to see if the transport properties degradeafter several runs.5 Another method to test the stability isto anneal the samples in a furnace at different temperaturesfor a varying amount of time and then measure their properties to study the effects of both the annealing time andtemperature.6–9 Both methods are useful for basic informationabout the thermal stability, however, both have shortcomings.Measurements taken at individual temperature intervals canmiss important information such as phase transitions. Andannealing samples in an oven at different temperatures fordifferent times is useful, but the question of how many different temperature or time intervals should be established isdifficult to answer. If there are not enough intervals, information may be missed or misinterpreted. If there are too many0034-6748/2012/83(11)/115114/7/ 30.00intervals, then a great deal of time and/or sample preparationis required in order to obtain accurate information.A solution to this problem can be made by measuringthe transport properties in situ as the material is being annealed and/or cycled. The difficulty in decoupling the parameters in ZT2 can now be seen as beneficial because by onlymeasuring one transport property, accurate information aboutthe material can be obtained as a function of both temperature and time; two important variables upon which thermalstability is dependent. We measure resistivity in situ as it isthe most accurate measurement among the transport properties contributing to ZT. In situ resistivity measurements arecommonly made on thin films to determine their temperature stability, but the authors have found no evidence or description of an experimental setup for in situ resistivity measurements on bulk thermoelectric materials to determine howproperties change as a function of both annealing time andtemperature. The Experimental details section describes thedeveloped setup to measure the resistivity of a bulk samplefrom room temperature to 550 C. It will be shown that accurate temperature control and continuous measurement allowfor much quicker and more accurate results for the determination of the thermal stability.The paper is divided into two sections. The first sectiondescribes in detail the experimental apparatus and configuration where standard materials are compared to ensure anaccurate resistivity measurement as a function of temperature.The second section uses in situ resistivity measurementsduring temperature cycling to demonstrate the quick and easyextraction of additional information on the thermal stabilityof a material. Finally, other potential uses for the setup arestated but these data are not included.EXPERIMENTAL DETAILSResistivity measurements can be made either using a fourpoint probe method on a bar shaped sample, or the Van derPauw (VdP) technique.11 The use of four probes negates any83, 115114-1 2012 American Institute of PhysicsDownloaded 09 Dec 2012 to 136.167.96.29. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights and permissions
115114-2Lukas et al.Rev. Sci. Instrum. 83, 115114 (2012)FIG. 1. Diagram of experimental setup for in situ resistivity measurements. Alumina rods, 1/4 in., (F) are connected to SS pieces (E) machined with hole sizesa fraction larger than the alumina rods. The sample (G), not drawn to scale, is mechanically pressed with a Ti screw (C) against the heating block (A) whichhouses a cartridge heater (B). Ni wire (H) is sparkwelded to the sample and mechanically connected (D) to external feedthroughs. The temperature is read by amechanically clamped K-type thermocouple (K). The inset shows the wiring configuration for a sample being measured using the VdP technique.concerns about contact resistance.12, 13 For thermoelectric materials alternating current (ac) is used to negate any voltagebuild up due to the Peltier and Seebeck effects.12 A frequencyof 13.7 Hz is used in the described setup, which is sufficientbecause the Peltier effect takes a finite amount of time to manifest itself which is usually on the order of 1 s.12The resistivity of samples is measured by the experimental setup (Figs. 1 and 2) in a helium backfilled vacuum chamber. The resistivity measurements must be performed in anoxygen free environment at elevated temperature to preventthe oxidation of the material, unless that is the purpose of themeasurements. Since the chamber must create an isolated environment, Viton o-rings are used to seal the chamber. Theo-rings have a maximum operating temperature which whenexceeded cause the o-rings to fail. In order to maintain an optimal operating temperature, a coolant plate (J) is used to dissipate excess heat. The coolant plate is kept at roughly 14 Cvia a closed loop water coolant system. This has proven to beenough cooling power to keep the o-rings under their maximum operating temperature, 100 C, while the inside of thechamber reaches temperatures in excess of 550 C. The chamber is evacuated with a mechanical pump down to pressuresof 10 Pa. The chamber is then backfilled with He gas, typically ambient pressure because it was determined that thecartridge heaters (B), which supply the heat, function betterwhen in the presence of an exchange gas as opposed to invacuum. Feedthroughs for electrical leads for current, voltage, heater power, and thermocouple inputs were mechanically fabricated. A sketch of the setup is shown in Figure 1.Nickel wire (3 mil) (H) is spark welded to each sample andthen mechanically connected (D) to copper wire leads whichCDAKBFIG. 2. Actual photograph of the side view schematic drawing shown inFig. 1. There is no sample on the heating block. The slots in the base areto screw into the coolant plate for stability.Downloaded 09 Dec 2012 to 136.167.96.29. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights and permissions
115114-3Lukas et al.in turn are mechanically connected to the vacuum chamberfeedthroughs leading out of the chamber where the instrumentation for data acquisition are attached. Ni wire is used because it does not diffuse into the sample as readily as Cu, Au,or Ag which is of concern at high temperature. Temperatureis read using 24 gage K-type thermocouple wire from OmegaEngineering Inc. which is mechanically attached to the heating block with a screw (K); the 24 gage wire should be thickenough to negate any effects of “green rot” on the positive element which is a problem in oxygen depleted environments.10Mechanical connections are used at higher temperatures because solder or other electrically conducting epoxies are moredifficult to use due to their lower operating temperatures. Heatis provided by a 120 V, 400 W cartridge heater (B) fromOmega with a length of 3 in. and a 3/8 in. diameter. The cartridge heater is placed into a 1 1 3 in. stainless steel (SS)block (A) with a hole size slightly larger than the diameter ofthe cartridge heater.In an ambient environment the cartridge heater resting inthe SS block can typically reach 700 C. In vacuum this temperature is much more difficult to reach, and it is found thatplacing oxygen-free high thermal conductivity Cu shimstockinside the hole of the SS block creates greater surface contact area so that the SS block can remove the heat away fromthe cartridge heater, allowing the heater to reach higher temperatures without electrically shorting. The use of braze asthermal contact material could be beneficial; however, its useis avoided so that the cartridge heater can be easily removedin the case of a short or failure. Another necessity is to backfill the chamber with an exchange gas which enhances heattransfer from the heater to the SS block. Combining these twomethods allows the temperature to easily be raised to 550 C,and if necessary to reach up to 600 C. The sample (G), whichis not drawn to scale as discussed later, sits on top of a 0.1 mmthick layer of mica which is on top the of the SS block providing electrical insulation but the mica is also thin enoughwhere it can be assumed that the sample temperature is thesame as that of the heating block. To ensure the sample isthermally connected to the heating block, it is mechanicallypressed down onto the block from above with a thin, 1/16 in.diameter, alumina rod (F) which fits in a tungsten screw (C).Enough force is applied to ensure good thermal contact without fracturing the sample. The rod must apply pressure so thatthe sample contacts the heating block evenly. Excellent thermal contact is imperative, as any temperature non-uniformitywill generate a Seebeck voltage thereby creating error in theresistivity measurement. Temperature inhomogeneity can bedue to heat loss from conduction through the rod, from convection due to the surrounding gas, and from radiation. Temperature inhomogeneity can also be caused by non-uniformheating which can manifest itself in thicker materials of lowthermal conductivity heated at a fast rate; the sample basecontacting the heating block will be much hotter than the topcreating and internal temperature difference. An accurate theoretical determination of heat loss is difficult. Therefore todetermine whether temperature differences due to heat lossand sample thickness are negligible, several measurementsof different materials are compared with both standards andcommercial equipment. All results agree within experimentalRev. Sci. Instrum. 83, 115114 (2012)error, described below, and so any effects on the measurementdue to temperature differentials are considered negligible. It isdetermined that for materials with thermal conductivity valuesbetween 1 and 5 W/m-K the geometry for bar shaped sampleswith dimensions 2 2 12 mm3 and for disk shaped samples with a thickness to diameter ratio of 1:11 are requiredfor accurate measurements. For the VdP technique, a smallersample thickness should provide more uniform temperature,though at the sacrifice of mechanical strength. It is importantin either geometry that the sample surfaces are flat to ensureuniform contact to the heating block.The temperature is read and controlled by a PXR 4 (PID)temperature controller from Fuji Electric to which both theheater and K-type thermocouple are connected. The PXR 4allows the rate at which the temperature is increased or decreased to be accurately controlled. Temperature is simultaneously read using a NI 9211 data acquisition system fromNational Instruments. The resistance is read using the 370 ACResistance Bridge from LakeShore which uses an (ac) with afrequency of 13.7 Hz. A LabVIEW program records the temperature (NI 9211), resistance (LS 370), time, and allows theuser to set the time interval at which data are recorded. Unless otherwise noted, data are recorded roughly once everysecond. There is again concern of temperature uniformity inthe sample when slewing the temperature, too fast a rate willlead to incorrect resistivity values, while too slow a rate is impractical. Again measurements are compared with standardsand commercial measurements. Also, measurements made atdiscrete temperatures are compared with measurements madeon the same sample while slewing the temperature. It is determined that a slew rate of 1 C/min is optimal for this systemand investigated samples.For a bar shaped sample the resistivity is obtained fromρ RA/L where R is the resistance, A is the cross sectionalarea, and L is the voltage lead separation. The placement ofvoltage leads always satisfies the ratio 2w Ls – L where wis the thickness of the sample, Ls is the length of the sample,and L is the voltage lead separation which ensures uniformityof the electric field, or one-dimensional current flow, at thevoltage leads.12The VdP technique can be used to measure a sampleof any arbitrary shape as long as the sample is flat and issingly connected, meaning it does not contain any holes.11, 14The resistivity using the VdP technique is given by theexpression11, 14, 15ρ tπ (R12,34 R23,41 )F.ln(2)2(1)Where R12,34 is defined as the current flowing between points1 and 2, while the voltage is read between points 3 and 4 (insetFigure 1), R23,41 has the current between points 2 and 3 withvoltage read between 1 and 4, t is the thickness of the sample,and F is a correction factor which can be solved graphicallyand is given by15 Fexp[ln(2)/F]Rr 1.(2) arccoshRr 1ln(2)2Where the ratio Rr R12,34 /R23,41 . Thermoelectric materials have no widely accepted standard at high temperatureDownloaded 09 Dec 2012 to 136.167.96.29. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights and permissions
Lukas et al.Where σ (R) is the standard deviation obtained from binningthe resistance values (R) at a given temperature, σ (L) is theuncertainty in the voltage lead separation (L), and σ (A) is theuncertainty in the cross sectional area (A). The error bars displayed for the VdP method are given by σVdP (R) 2σ (t) 2σ (F ) 2σVdP (ρ) .(4) ρRtFWhere σ (t) is the uncertainty in the thickness of the sample(t), σ (F) is the uncertainty in the graphical determination ofthe correction factor (F) which we take to not exceed 3%, and (5)σVdP (R) σ (R12,34 )2 σ (R23,41 )2 .Where σ (R12,34 ) and σ (R23,41 ) are the standard deviation ofbinned resistance values for a given temperature, typical binsize is every 1 . It should be noted that Eq. (1) is writtenunder the assumption that the size of the contact points areinfinitesimal and the contacts are made directly on the edgeof the specimen. In reality, the wire will always have somefinite thickness and it is not possible to place the wire exactly on the edge of the sample, leading to additional error.This error is difficult to quantify but should not be exceed 2%as long as care is taken in wire placement.18 Therefore, it isnot taken into account in the expression for the error given inEq. (4).Figure 3 shows the resistivity data for constantan whilethe temperature is increased at discrete intervals, whileFigure 4 plots the dependence of the resistivity of nickelmeasured continuously over the temperature range. Constantan is measured at discrete temperatures on a bar shapedsample of dimensions 2 2 14 mm3 . The resistance valuesat each temperature are binned which gives the value ofσ (R) in Eq. (3). It can be seen that the data measured bythe constructed setup matches within 1% of the data takenby the ZEM-3 (ULVAC) on the standard constantan barprovided by ULVAC. The resistivity data of nickel (Fig. 4)are measured on a flat square shaped sample of dimensions16 16 2 mm3 using the VdP technique. The temperatureis increased continuously from 20 C – to 550 C at a rate of1 C/min to measure R12,34 . The sample is then cooled andwires reconfigured to measure R23,41 . The sample is againmeasured, while the temperature is increased at 1 C/min.Resistance values are binned every degree to obtain thestandard deviation. The experiments are performed fromroom temperature to up to 550 C. The agreement with theliterature data, we take their uncertainty as 7%,19 is within theexperimental uncertainty of the two measurement systemswith the absolute deviation at higher temperatures never6.52.0Percent Diffeerence (100*[ZEM-SSFPP]/ZEM)6.0-7-m)Four pointZEMResistivRvity (x10(NIST only recently developed a low temperature standard16 ),thus it is imperative to accurately understand and accountfor any sources of error in the measurement so that data canbe more accurately understood and communicated among research groups. The error bars for bar shaped samples from thepropagation of independent errors are given by17 σ (R) 2σ (ρ)σ (L) 2σ (A) 2 .(3)ρRLARev. Sci. Instrum. 83, 115114 ret( C)5.04.50100200300oTemperature ( C)FIG. 3. The resistivity of a standard bar-shape constantan sample is plotted versus temperature along with the uncertainty calculated from Eq. (3)demonstrating agreement between both systems. The inset plots the percentdifference in the resistivity between the developed setup and the commercially available ZEM-3.exceeding 6%.19 The ferromagnetic transition temperature isin very good agreement with the literature.20 The transitiontakes place at 355.5 C according to the literature,20 while themeasurement here gives a transition temperature of 354 Cwhich is well within the industrial error of 0.75% given forK-type thermocouples by Omega Engineering Inc.Nickel and constantan are standard metals, but both havelower Seebeck coefficients and higher thermal conductivitiesthan commonly used thermoelectrics. Therefore, several otherthermoelectric samples were run in order to validate the accuracy of the machine. These data are included in a supplementary material;21 however, the results are summarized. Forbar shaped samples measured in both the developed setup aswell as the ZEM-3, disagreement never exceeds 3% whichis within the experimental uncertainty of the above system.4.0x10-7Burkov 99.98% NiVDP 99.9993% NiResistivity ( 00600oTemperature ( C)FIG. 4. Resistivity of nickel is plotted against temperature. Measurementswere made using the VdP technique on a sample of 99.9993% purity fromAJA International, Inc. Values obtained from Ref. 19 for Ni of 99.98% purityare shown for comparison. Agreement is seen between the two data sets validating the system. The slight deviation at higher temperatures could be dueto the difference in sample purity.Downloaded 09 Dec 2012 to 136.167.96.29. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights and permissions
Lukas et al.RESULTSTo demonstrate the developed experimental apparatus,the resistivities of two thermoelectric material samples are2.0x10-5450 oC warming450 oCcoolingm)The absolute difference for VdP measurements never exceeds9%, which is still within experimental error when the uncertainty of 3% of the ZEM-3 is taken into account. There aretwo probable reasons for a greater disagreement in the VdPmeasurements. The first is that the resistivity is being compared between two different samples. One is a thin disk usedfor VdP measurements, while the other is a bar that is not cutfrom the same disk; the ZEM can only measure bar shapedsamples. There can be slight variations among transport measurements of different samples of the same TE material. Thesecond reason is due to the aforementioned effects of finitecontact size and probe placement near the edge, so the difference noted above is not unexpected.The apparatus has been designed and benchmarkedto study thermoelectric materials, in particular, bar shapedsamples of dimensions 2 2 12 mm3 and disk shapedsamples of 12 mm diameter and 1 mm thickness. Cautionand care should be taken if measuring samples of alternatedimensions and/or non-thermoelectric materials. As mentioned previously, samples that are too thick can have internaltemperature gradients that lead to additional voltages therebyaltering the measurement. The other potential cause of atemperature gradient could come from heat loss through radiation, convection due to the surrounding gas, or conductionthrough the alumina rod. While multiple comparisons madebetween standards, commercial equipment, and the describedapparatus demonstrate any effect outside of statistical erroris negligible, accuracy could be affected by varying thesample geometry or for non-thermoelectrics. A heat shieldsurrounding the sample along with narrowing the contactingcross sectional area of the alumina rod could be introducedas improvements if systematic errors are discovered. Anotherpotential improvement could be made by introducing aconstant force spring as the method of applying force downon the sample instead of the tungsten screw. With the spring,any change in sample thickness due to thermal expansionor softening would not affect the contacting force and somaintain uniform contact with the heating block; this is notof concern in our samples as any thermoelectric materials ofinterest for commercial use need to have stable mechanicalproperties and, therefore, any material that expands or softenswould not be systematically investigated. The described setuphas also been used to measure small, 1 mm3 , single crystalsamples. Due to the sample size 1 mil Pt wire was used andthe sample was thermally connected to the heating blockby embedding the sample in a high temperature dielectricepoxy (Omega 600 from Omega Engineering Inc.). Slighterror could be introduced in this instance from the differencebetween the sample temperature and the measured temperature. However, by heating the block at a slower ramp ratethe difference between the sample and measured temperatureshould not exceed a few degrees. Work is currently underwayto improve and benchmark the apparatus and accuratelydetermine the error for smaller single crystal samples.Rev. Sci. Instrum. 83, 115114 (2012)1.5x10-5Ressistivity (115114-5200 oCwarming/cooling1.0x10-5100200300400oTemperature ( C)FIG. 5. Resistivity is plotted against temperature for two temperature cyclesfor Cu0.01 Bi2 Te2.7 Se0.3 hot pressed at 450 C. Negligible change is seen inthe material, while the temperature remains below the pressing temperature.measured during in situ annealing and/or temperature cycling.Figure 5 shows raw, not binned, data for a Cu0.01 Bi2 Te2.7 Se0.3sample that has been temperature cycled. The sample is prepared via ball milling and dc hot pressing techniques described previously;5 the sample in Figure 5 is hot pressedat 450 C. The temperature is incrementally cycled by running from 40 C up to 200 C at a rate of 5 C/min, then from200 C back to 40 C at a rate of 5 C/min. The temperatureis then ramped from 40 C to 225 C and then from 225 Cback to 40 C. The system remains at each maximum temperature for 10 min before cooling back down. This procedure is continued while increasing the maximum temperatureeach time by 25 C up to a temperature of 450 C. A ramprate of 5 C/min is used to expedite the measurement time,though there is a slight error introduced due to the higher ramprate, the difference between the warming and cooling curvesfor the 200 C run is seen to fall on top of each other indicating a negligible difference from the ramp rate. It can beseen that the sample exhibits metallic-like behavior. Whileseveral different runs were recorded, only the first run to atemperature of 200 C and the final run to 450 C are shownfor clarity; the data from all intermediate temperatures lie inbetween the warming curves for both 200 C and 450 C. Thevalues for the resistivity change by less than 5%, while thetemperature remains below the pressing temperature. However, once the pressing temperature is reached and slightlyexceeded due to the overshooting of the system temperatureat large ramp rates, the resistivity value is lowered by about13% from its initial value. The reason is that the dc hot pressmethod essentially anneals the sample at the pressing temperature, and if the pressing temperature is exceeded duringmeasurement or operation, there are irreversible changes tothe transport properties of the material due to further annealing. While both Seebeck coefficient and thermal conductivityvalues are just as important as resistivity values for overalloptimization, it is evident how quickly information about annealing time, temperature and phase transitions can be ascertained from the continuous measurement capability.Downloaded 09 Dec 2012 to 136.167.96.29. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights and permissions
Lukas et al.Rev. Sci. Instrum. 83, 115114 (2012)400300WarmingCooling-67.0x10200Resistivity ( --m)RResistivity ( -m)oTemperature ( 00200300400oTemperature ( C)Time (A.U)(b)(a)FIG. 6. (a) Resistivity and temperature are plotted against time. (b) The resistivity is plotted with temperature showing negligible difference in the warming andcooling curves, even at different cooling rates.Another possibility that is easy to realize besides temperature cycling is in situ annealing measurements as a functionof time, which is measured for Yb0.35 Co4 Sb12 (Fig. 6). Thesample can be brought to a specific temperature and remain atthat temperature to study the effects of annealing or operatingtemperature over a period of time. The typical hot junctiontemperature range of operation for skutterudites is between400 C and 500 C for waste heat applications.13 The sampleis heated from 20 C to 400 C at a rate of 5 C/min (Fig. 6).The temperature is then held at 400 C for 36 h. Then thetemperature is lowered from 400 C to 20 C again at a rateof 5 C/min. As mentioned previously the time interval atwhich data are recorded can be changed. During warmingand cooling the data are recorded roughly every second. Tominimize the number of data points during the 36 h period,data are recorded every 10 min. The recording time for thedata can be adjusted in a wide range, commonly between1 and 3600 s for the developed setup. In the experiments,the ramp rate is constant on the way up as expected and thetemperature is stable at 400 C for the entire 36 h. The resistivity at 400 C changes by less than 1% over the 36 h period(Fig. 6(a) bottom). The cooling rate is constant until atemperature of 80 C is reached and the system does not havethe ability to cool at the 5 C/min through convective andconductive cooling (Fig. 6(a) top). The resistivity shows nohysteresis even with the different cooling rates below 80 C,the difference in the room temperature values of the resistivitybefore and after heating is about 1%. The data sets show bothcompounds are thermally stable over the studied
An experimental setup for determining the electrical resistivity of several types of thermoelectric materials over the temperature range 20 T 550 C is described in detail. One resistivity measure-ment during temperature cycling is performed and explained for Cu 0.01Bi 2Te 2.7Se 0.3 and a second measurement is made on Yb 0.35Co 4Sb
factor of organic thermoelectric materials is three orders of magnitude lower than that of conventional inorganic thermoelectric materials [58-60]. To overcome these restrictions, organic polymer-inorganic thermoelectric composite structures have been developed, and satisfactory thermoelectric performance has been obtained over time.
A thermoelectric generator converts heat directly into electricity. As heat moves from a gas burner through a thermoelectric module, it causes an electrical current to fl ow. The heart of a Global thermoelectric generator is a hermetically sealed thermoelectric module (thermopile) which
Such metalloid thermoelectric materials would allow for greater thermoelectric efficiency than the more robust oxide thermoelectric materials required by the 1 st and 2 nd generation receivers. Activities within Task 3 related to the dev
energy into electricity by using thermoelectric transformation materials. A thermoelectric power converter has no moving parts, and . Obviously, Module 4, with the cost of 3.4 dollar each, has the highest power-cost ratio. . is the efficiency of the thermoelectric module, P is the output
thermoelectric generator devices made of segmented thermoelectric elements”, ICT 2010 160 mm Water manifold Hot oil manifold TE Circuit (Bi 2 Te 3 subassemblies) [1] “Performance Results of a High Power Density Thermoelectric Generator: Beyond the Couple”, ICT 2008 At Gentherm, c
Champier et al., Studied a thermoelectric incorporated in a multifunction wood stove and concluded that adding TE (thermoelectric) generators can provide electricity up to 9.5 W. [2]Cheng-Ting Hsu, Gia-Yeh Huang, et al., [2011], Using a network of 24 thermoelectric
Thermoelectric Cooler Fig.2.Thermoelectric Cooler. V1- DC power supply. The Cold side is heat-insulated. The Hot side is located on Heat sink. RqR – Heat Sink (CPU cooler). Vte – Environment Temperature { C}. a) Temperature versus Voltage b) Current versus Voltage Fig.3. TEC1-12706 characteristics. Thermoelectric Cooler Mode.
Thermoelectric Module Heat Sink Thermoelectric Module Heat Sink Voids lled with air act as an insulator Heat can only pass through where the surface actually touches Figure 4 shows how the remaining valleys create voids through which heat energy can barely pass through, in e ect creating an insulated area – not the ideal thermal interface.
