Academia.eduAcademia.edu

Figure 2 – uploaded by F. Edler

See full PDFdownloadDownload figure
Figure 2: Measurement routine for the determination of the Seebeck coefficient. At first heater1 is switched on and a temperature gradient arises. After a short time the heater is switched off and the system relaxes; finally the step is repeated with heater2. Temperatures and voltages are recorded for the complete measurement cycle, but only the data aquired during relaxation is used for the calculation of S.

Figure 2 Measurement routine for the determination of the Seebeck coefficient. At first heater1 is switched on and a temperature gradient arises. After a short time the heater is switched off and the system relaxes; finally the step is repeated with heater2. Temperatures and voltages are recorded for the complete measurement cycle, but only the data aquired during relaxation is used for the calculation of S.

Related Figures (9)

Figure 1: Sample holder for o and S measurement. Two gradient heaters can be used to create < temperature gradient across the sample which results in Seebeck voltages. The scheme in a) displays the temperature profile during a measurement of the Seebeck coefficient. The photograph in b) shows the sample holder with a mounted sample, while the scheme in c) illustrates the mechanical and electrica connections (only left half shown for clarity). The sheathed thermocouples are used to measure the sample temperatures and the thermoelectric voltages; from these the Seebeck coefficient S of the sample under test can be calculated. For the measurement of o a current is driven through the sample using the thermocouples for current supply. Additional voltage sensors in line with the thermocouples are used tc record the Ohmic voltage drop. Seebeck coefficient since otherwise thermal contact resistances will lead to ar erroneous temperature reading and thus an incorrect result for S. The sample itself is pressed onto the sample holder using a headless screw and a “T”-piece that can be plugged into a hole drilled into the headless screw. The head of the “T’-piece is rotatable and can therefore also be used for samples with non-parallel surfaces. rotatable and can therefore also be used for samples with non-parallel surfaces. plugged into a hole drilled into the headless screw. The head of the “T”-piece is
Figure 1: Sample holder for o and S measurement. Two gradient heaters can be used to create < temperature gradient across the sample which results in Seebeck voltages. The scheme in a) displays the temperature profile during a measurement of the Seebeck coefficient. The photograph in b) shows the sample holder with a mounted sample, while the scheme in c) illustrates the mechanical and electrica connections (only left half shown for clarity). The sheathed thermocouples are used to measure the sample temperatures and the thermoelectric voltages; from these the Seebeck coefficient S of the sample under test can be calculated. For the measurement of o a current is driven through the sample using the thermocouples for current supply. Additional voltage sensors in line with the thermocouples are used tc record the Ohmic voltage drop. Seebeck coefficient since otherwise thermal contact resistances will lead to ar erroneous temperature reading and thus an incorrect result for S. The sample itself is pressed onto the sample holder using a headless screw and a “T”-piece that can be plugged into a hole drilled into the headless screw. The head of the “T’-piece is rotatable and can therefore also be used for samples with non-parallel surfaces. rotatable and can therefore also be used for samples with non-parallel surfaces. plugged into a hole drilled into the headless screw. The head of the “T”-piece is
Figure 3: Plot of Ug vs. U,. The slope of Ug vs. U,is used to determine the Seebeck coefficient, see Eq. (7). The linear fit is excellent as can be deduced from the correlation coefficient R. Measurement uncertainty analysis and comparison with reference data
Figure 3: Plot of Ug vs. U,. The slope of Ug vs. U,is used to determine the Seebeck coefficient, see Eq. (7). The linear fit is excellent as can be deduced from the correlation coefficient R. Measurement uncertainty analysis and comparison with reference data
Figure 4: Comparison of the Seebeck coefficient from the NIST SRM 3451 with the custom-built system CTEM (blue). The black points indicate the published data from NIST, while the error bars indicate the stated measurement uncertainty. The black and the blue curve show excellent agreement with maximal deviations of 1%.
Figure 4: Comparison of the Seebeck coefficient from the NIST SRM 3451 with the custom-built system CTEM (blue). The black points indicate the published data from NIST, while the error bars indicate the stated measurement uncertainty. The black and the blue curve show excellent agreement with maximal deviations of 1%.
Figure 5: Comparison of the measurement results for the Seebeck coefficient between the CTEM and the HTSo. A hot-pressed skutterudite was used as sample. Both data sets show good agreement. After proofing accurateness of the CTEM we can now in a second step compare the measurement results between CTEM and HTSg. Results for a CoSb3 skutterudite sample are plotted in Figure 5 and show very good agreement between CTEM and HTSo results up to ~ 550K which is the current maximum operation temperature for the CTEM- setup.
Figure 5: Comparison of the measurement results for the Seebeck coefficient between the CTEM and the HTSo. A hot-pressed skutterudite was used as sample. Both data sets show good agreement. After proofing accurateness of the CTEM we can now in a second step compare the measurement results between CTEM and HTSg. Results for a CoSb3 skutterudite sample are plotted in Figure 5 and show very good agreement between CTEM and HTSo results up to ~ 550K which is the current maximum operation temperature for the CTEM- setup.
Figure 6: Comparison of Seebeck coefficient of Cu,;Ni;, measured with three different setups: HTSo and CTEM in-house as well as an apparatus from the PTB. The agreement between all three data sets is very good.
Figure 6: Comparison of Seebeck coefficient of Cu,;Ni;, measured with three different setups: HTSo and CTEM in-house as well as an apparatus from the PTB. The agreement between all three data sets is very good.
Figure 7: Seebeck coefficient and electrical conductivity of Cus,Ni,,Mn,. Also shown is the comparative measurement of the Seebeck coefficient of the PTB which again confirms the measurement data of the HTSo. The results for the second sample (CusgNiggMniz, 24 mm*11 mm*0.3 mm, ISOTAN fron Isabellenhtitte Heusler GmbH & Co. KG, Germany) are shown in Figure 7 together wit! the measurement results for the electrical conductivity. The magnitude of the sample’ Seebeck coefficient is around —37 uV/K at room and increases with temperature. Th data sets for both samples compare very well and show a maximum deviation of < 49 from each other. The electrical conductivity of the second sample shows wea temperature dependence. This is not surprising as the alloy has a similar composition a constantan.
Figure 7: Seebeck coefficient and electrical conductivity of Cus,Ni,,Mn,. Also shown is the comparative measurement of the Seebeck coefficient of the PTB which again confirms the measurement data of the HTSo. The results for the second sample (CusgNiggMniz, 24 mm*11 mm*0.3 mm, ISOTAN fron Isabellenhtitte Heusler GmbH & Co. KG, Germany) are shown in Figure 7 together wit! the measurement results for the electrical conductivity. The magnitude of the sample’ Seebeck coefficient is around —37 uV/K at room and increases with temperature. Th data sets for both samples compare very well and show a maximum deviation of < 49 from each other. The electrical conductivity of the second sample shows wea temperature dependence. This is not surprising as the alloy has a similar composition a constantan.
Figure 8: Spatial scan of the Seebeck coefficient of a B-FeSi, sample. It can be seen that the sample is very homogeneous; the full width half maximum of the Seebeck coefficient distribution is only 1.7% of the mean value.
Figure 8: Spatial scan of the Seebeck coefficient of a B-FeSi, sample. It can be seen that the sample is very homogeneous; the full width half maximum of the Seebeck coefficient distribution is only 1.7% of the mean value.
Figure 9: Thermal cycling program (a) and corresponding results for the Seebeck coefficient. In b) the corresponding raw measurement data is shown together with polynomial fits for better clarity. No significant change of the Seebeck coefficient can be seen after repeated thermal cycling of the sample, partially above 1000 K. We have investigated the thermal stability of FeSi. samples employing the presented setup. We have therefore exposed one of the samples to several thermal cycles up to 1000 K as shown in Figure 9a. The corresponding results for the Seebeck coefficient are presented in Figure 9b. For clarity polynomial fits of the measurement data are plotted together with the raw measurement results. The maximum deviation of the polynomial fits from each other is = 2% of the absolute value and shows no systematic drift with temperature. We can therefore deduce that the prepared FeSi, is thermally stable up to 1000 K, which is one crucial requirement for a high temperature standard material. 1000 K, which is one crucial requirement for a high temperature standard material.
Figure 9: Thermal cycling program (a) and corresponding results for the Seebeck coefficient. In b) the corresponding raw measurement data is shown together with polynomial fits for better clarity. No significant change of the Seebeck coefficient can be seen after repeated thermal cycling of the sample, partially above 1000 K. We have investigated the thermal stability of FeSi. samples employing the presented setup. We have therefore exposed one of the samples to several thermal cycles up to 1000 K as shown in Figure 9a. The corresponding results for the Seebeck coefficient are presented in Figure 9b. For clarity polynomial fits of the measurement data are plotted together with the raw measurement results. The maximum deviation of the polynomial fits from each other is = 2% of the absolute value and shows no systematic drift with temperature. We can therefore deduce that the prepared FeSi, is thermally stable up to 1000 K, which is one crucial requirement for a high temperature standard material. 1000 K, which is one crucial requirement for a high temperature standard material.
Figure 10: Comparison of the Seebeck coefficient of two FeSi, samples. The two samples show very similar values which indicate a reproducible sample fabrication. Another important criterion for a reference material is the reproducibility of th samples. Figure 10 compares the measurement results of the first sample with a furthe sample that is not from the same batch but fabricated under the same conditions. Th measurement results are very close to each other as exemplified by the polynomial fit that are plotted for clarity. These fits differ by less than 2% over the whole temperatur range, which is less than the measurement uncertainty. Comparison of two samples i clearly not sufficient to prove reproducibility, but it indicates that the fabrication proces is well understood and a consistent production of samples is feasible. is well understood and a consistent production of samples is feasible.
Figure 10: Comparison of the Seebeck coefficient of two FeSi, samples. The two samples show very similar values which indicate a reproducible sample fabrication. Another important criterion for a reference material is the reproducibility of th samples. Figure 10 compares the measurement results of the first sample with a furthe sample that is not from the same batch but fabricated under the same conditions. Th measurement results are very close to each other as exemplified by the polynomial fit that are plotted for clarity. These fits differ by less than 2% over the whole temperatur range, which is less than the measurement uncertainty. Comparison of two samples i clearly not sufficient to prove reproducibility, but it indicates that the fabrication proces is well understood and a consistent production of samples is feasible. is well understood and a consistent production of samples is feasible.
Related topics:
Electronic MaterialsElectrical and Electronic Engineering
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved