A setup for Seebeck coefficient measurement through controlled heat pulses
A setup is designed for measuring the Seebeck coefficient of materials in form of thin film, bar and wire. The main feature of this setup is control in heating and cooling cycles for the studies like phase transitions. In this setup heat pulse is used to generate the temperature gradient using the same temperature controller, no extra current source is required. The uncertainty in Seebeck coefficient measurement is found to be while temperature stability is (at 320 K), confirmed over time at fixed temperature as well as during temperature ramping. To demonstrate the capabilities of this setup, temperature dependent Seebeck coefficient of standard wire samples such as Au-Fe (0.07 %), chromel, platinum and thin films of Pt and F doped SnO are presented. We have tested the setup in temperature range 100 K to 320 K, while it does not have any intrinsic limitation in going down to liquid He temperatures. For higher temperatures above 320 K limitation is due to gluing material like varnish.
Currently, a larger Physics community among those working in the direction to find new energy resources are engaged particularly in waste heat management and conversion of waste heat into electricity. The urgency for thermoelectric materials that can convert heat into electricity is high, because of global warming and limited natural energy sources. Values of the Seebeck coefficient provides direct initial allusion for possible application of materials. For an effective thermoelectric material the correlation between the Seebeck coefficient (S), electrical conductivity () and thermal conductivity () plays an important role Dresselhaus et al. (2007); Liu et al. (2015). These all together decides the efficiency of the material for practical use through the relation Snyder and Toberer (2008) ZT = ST/, where ZT is figure of merit. Apart form this, the measurement of the Seebeck coefficient is useful to study the nature of mobile charge carriers (electron or holes), by the sign of S. It is good tool also for the studies of phase transitions, idea of density of states and electron-magnon or electron-phonon interactions Ahad et al. (2017); Ziman (1979) etc.
Development of the Seebeck coefficient (S) measurement system has always been resurgent among the researchers. Several efforts have been made to design efficient, cost effective and accurate setup Tripathi, Bala, and Asokan (2014); Mulla and Rabinal (2016); Boffoue et al. (2005); Mun et al. (2010); Wang et al. (2018); Sharath Chandra et al. (2008); Soni and Okram (2008); Sarath Kumar and Kasiviswanathan (2008). In the setup of Tripathi Tripathi, Bala, and Asokan (2014) they claim uncertainty in the measurement of S 0.5 , while the sample mounting is convenient and their setup requires a nano-voltmeter, a temperature controller and a current source. Soni Soni and Okram (2008) claim fast measurements but their setup requires two separate temperature controllers. Sharath Sharath Chandra et al. (2008) reported almost similar arrangement like Soni in PPMS (physical properties measurement system) probe. In the design of Jonathan DâAngelo, Downey, and Hogan (2010) they are able to measure from 300 K to 1273 K and the absolute error of measurement was 3-5 %. E. Mun Mun et al. (2010) claim uncertainty in temperature range 2 to 350 K. Some reports focus on the sample mounting procedure, like sandwich mounting Mulla and Rabinal (2016); Boffoue et al. (2005), etc. The bridging mounting has been popular because of its suitability for thin films as well as bar or wire Mun et al. (2010); Ravichandran et al. (2011); Tripathi, Bala, and Asokan (2014); Fu et al. (2017); Kedia, Singh, and Chaudhary (2018). A recent review on the Seebeck coefficient measurement setup has mentioned all the merits and demerits of previously reported setups and conclude that one has to analyze the measurement error in their setup Wang et al. (2018) for the tracking of design demerit.
In our setup we have used the bridging mounting and small for negligible influence on the base temperature. We apply controlled heat pulses on one end to produce periodic . For estimation of S we use the linear fit method. Fitting procedure reduces the effect of spurious or offset voltages Wang et al. (2018). We have attempted to sum-up previous methods with motive of less use of instruments (only a temperature controller and a nano-voltmeter). We show excellent agreement between measured and reported data of standard materials. We also show the uncertainty in S () with time as well as temperature which is an essential parameter for reproducibility Yazdani, Kim, and Pettes (2018).
Ii Experimental Setup
In the following subsections we describe the design of sample holder, measurement strategy, accuracy, reproducibility, limitations and flexibility of the setup. The setup is automated with the National instruments LabView interface development Elliott et al. (2007) for the operation and data collection.
ii.1 Design of sample holder
Fig. 1 (upper panel) displays a photograph of an optimized sample holder for producing effective temperature gradient (). Now we describe the individual components in the sample holder, which are marked from 1 to 9. We have used oxygen free highly conducting (OFHC) copper bars . Each copper bar is 50 mm long, 6 mm wide and 2 mm thick. We have inserted a piece of 6 mm x 4 mm x 2 mm copper in between the bars as a heat exchanger . At the bottom, a 50 Ohm heater of insulated manganin wire has been mounted to include both bars for the base temperature. Two separate Pt100 sensors have been employed to measure the temperature of each bar. A SMD thermistor of 100 Ohm is glued on one of the bar through cigarette paper. On both sides of the bars small pieces (6 mm X 4mm) of copper foils were glued through the cigarette paper , to make electrical insulation. Each of the copper foil piece has been mechanically anchored with the copper wire of high gauge for measuring voltages (V+/V-). To measure temperature difference between the bars, on one side we have thermally anchored the differential thermocouple of chromel-copper. This arrangement will provide identical temperature environment to both, the thermocouple and the sample under measurement. Lower panel of Fig. 1 shows the schematic of sample holder with the distribution of temperature on the bars during the measurement. All the wires from the sample holder have been soldered on printed circuit board (PCB) which is glued on the hylam socket. This whole arrangement is screwed with a brass at the bottom of the insert which is made of seamless stainless steel tube. All the wires from PCB to instruments are connected through a brass vacuum feedthrough. The sample holder assembly has been covered with brass jacket for vacuum. At the time of measurement the probe will have to hang within liquid nitrogen dewar and the sample space is evacuated. The wires coming from the sample holder are connected to temperature controller (Cryo-con 22C) and Keithley 2182A nano-voltmeter.
ii.2 Strategy for the temperature gradient , measurement of differential voltage and protocol
This setup uses only one heater for the base temperature and requires only one temperature controller. This heater is sufficient to heat the sample upto 400 K with the limitation of insulation of manganin heater wire. This heater can be ramped with the controlled ramp using one of the two loops of temperature controller. It makes the temperature of both the bars identical. Bar with SMD thermistor is connected with the second control loop of temperature controller and can be controlled in pulse mode. By controlling the output power and optimizing the time for the pulse to be ON, one may get the desirable of about 2 K to 5 K. We optimized that by applying 3 % power to thermistor for 50 seconds, of about 2 K can be achieved. Fig. 2 (a) shows the profile of with time.
Our measurement program first stabilizes the set (desired) temperature and 60 sec after stabilization a heat pulse is applied (see Fig. 2 (c)). Heat pulse raises the temperature of the bar with SMD and a of desired value is achieved. Once desired is achieved, SMD is kept unpowered until approaches close to zero. Simultaneously of the sample is measured for each (see Fig. 2 (d)). In Keithley 2182A nanovoltmeter, one channel measures the voltage from the sample while the other channel is dedicated to thermocouple. Time duration when decreases and approaches towards zero has been optimized for the fitting and data collection for maximum accuracy. This process provides S value for the base temperatures. The temperature increment/decrement is in the ramp mode throughout the measurement. We have used 0.3 K/min for the standard samples. The advantage of our setup is the control in the heating and cooling cycle both, which is very important for studies of the first order phase transition. Fig. 2 (b) shows the modulation of heat pulses on base temperature in heating and cooling cycles. We have measured the Au-Fe (0.07%) and chromel in heating-cooling cycles and excellent overlap of both the cycles is observed (see Fig. 3 (c) and Fig. 3 (d)).
ii.3 Accuracy and data reproducibility
In this section we will describe about quality of the data. Usually, the main error in Seebeck measurement comes from the error in the measurement of . To overcome this we used the thermocouple S data and measured voltage difference () to get temperature difference () through the relation Tripathi, Bala, and Asokan (2014); Fu et al. (2017) of S and , Eq. 1. This method has been proven to reduce the error Ramu et al. (2012) in . Using value of and simultaneously coming from the sample, one can calculate the value of S. For maximum accuracy, we perform linear fit (see Fig. 2 (e)) of and during the measurement. To check the stability of data, we have made four consecutive measurements in heating and cooling cycles (see Fig. 3 (d)). S measurements from different cycles overlap very well.
In the following, we will discuss about the flexibility and limitation of the setup. Fig. 3 (e) schematically shows the modulation of heat pulse on a stable (left) and a ramping (right) base temperature with ramp rate m (= dT/dt). One can see that the ramp rate (m) affects to the effective fitting range (). For different ramp rates, we can easily change range by observing the linearity of ( vs ) curve (Fig. 2 (e)). However, the maximum rate of measurement (m) should not exceed the change in () within the fitting range () . This flexibility allows us to successfully measure the S value upto 2 K/min with good overlap (see Fig. 3 (f)). Faster measurement speed is suitable only if there are no phase transition, however, to observe a phase transition ramp rate must be kept slower.
In this section we will show the measured data of high purity wires of platinum, chromel and iron doped gold (0.07 %) of high gauges. All these wires have been bridge mounted using conducting silver paint (RS 186-3600, RS components, UK). The use of silver paint is very convenient and one may easily remove it by amyl acetate. After curing the silver paint under the heating lamp, one can start the measurement. Fig. 3 (a) and (b) show the data of a chromel wire sample. The uncertainty in measured S is and stability in base temperature is (at 320 K) over a large measurement time ( 10 sec). Stability for such large time durations are not reported in literature. Fig. 4 (a) shows the measured Seebeck coefficient (S) of chromel vs copper along with the reported data Tripathi, Bala, and Asokan (2014). We observe excellent agreement with the literature. Measured data of the pure platinum wire of very high gauge ( 30 ) also matches well with the earlier reports Burkov et al. (2001); Tripathi, Bala, and Asokan (2014) (see Fig. 4 (b)). The crossover of sign change in S for platinum also confirms the accuracy of sample temperature. In order to show the suitability of our setup for the thin film Seebeck measurement, data of a 58 nm thin Platinum film on quartz substrate (deposited by magnetron sputtering) and commercially available (Techinstro, India) F doped SnO (thickness 200 nm) on soda lime glass are shown in the Fig. 4 ((c) and (d)). These data confirm the suitability of setup for the Seebeck coefficient measurements of thin films.
We have presented an effective automated design for the Seebeck coefficient measurement for samples in form of wire, bar and thin films. Controlled heat pulse is employed to produce a temperature gradient using the same temperature controller and simultaneous measurement allows estimation of S through linear fit. The setup is able to reproduce the data in heating and cooling cycles, hence, it is suitable for study of the phase transitions. S measurement shows uncertainty upto and temperature stability (at 320 K). Measurement of S taken at ramp rate upto 2 K/min are shown to be overlapping with data acquired with slower rate (0.3 K/min).
Acknowledgements.We are thankful to G. S. Okram for providing high purity chromel and Au-Fe (0.07 %) wires, R. Rawat for providing pure platinum wire and M. Gupta for providing a Pt thin film. One of the authors AA acknowledges UGC, New Delhi, India for financial support in the form of MANF (2016-17/MANF-2015-17-UTT-53853) scheme. DKS acknowledges support from DST-New Delhi, India through grant no. INI/RUS/RFBR/P-269.
- Dresselhaus et al. (2007) M. S. Dresselhaus, G. Chen, M. Y. Tang, R. Yang, H. Lee, D. Wang, Z. Ren, J.-P. Fleurial, and P. Gogna, Adv. Mater. 19, 1043 (2007).
- Liu et al. (2015) W. Liu, Q. Jie, H. S. Kim, and Z. Ren, Acta Mater. 87, 357 (2015).
- Snyder and Toberer (2008) G. J. Snyder and E. S. Toberer, Nat. Mater. 7, 105 (2008).
- Ahad et al. (2017) A. Ahad, D. K. Shukla, F. Rahman, S. Majid, Tarachand, G. S. Okram, and D. M. Phase, Acta Mater. 135, 233 (2017).
- Ziman (1979) J. M. Ziman, Principles of the Theory of Solids (Cambridge university press, 1979).
- Tripathi, Bala, and Asokan (2014) T. Tripathi, M. Bala, and K. Asokan, Rev. Sci. Instrum. 85, 085115 (2014).
- Mulla and Rabinal (2016) R. Mulla and M. Rabinal, ACS Comb. Sci. 18, 177 (2016).
- Boffoue et al. (2005) O. Boffoue, A. Jacquot, A. Dauscher, B. Lenoir, and M. Stölzer, Rev. Sci. Instrum. 76, 053907 (2005).
- Mun et al. (2010) E. Mun, S. L. Bud’ko, M. S. Torikachvili, and P. C. Canfield, Meas. Sci. Technol. 21, 055104 (2010).
- Wang et al. (2018) C. Wang, F. Chen, K. Sun, R. Chen, M. Li, X. Zhou, Y. Sun, D. Chen, and G. Wang, Rev. Sci. Instrum. 89, 101501 (2018).
- Sharath Chandra et al. (2008) L. Sharath Chandra, A. Lakhani, D. Jain, S. Pandya, P. Vishwakarma, M. Gangrade, and V. Ganesan, Rev. Sci. Instrum. 79, 103907 (2008).
- Soni and Okram (2008) A. Soni and G. Okram, Rev. Sci. Instrum. 79, 125103 (2008).
- Sarath Kumar and Kasiviswanathan (2008) S. Sarath Kumar and S. Kasiviswanathan, Rev. Sci. Instrum. 79, 024302 (2008).
- DâAngelo, Downey, and Hogan (2010) J. DâAngelo, A. Downey, and T. Hogan, Rev. Sci. Instrum. 81, 075107 (2010).
- Ravichandran et al. (2011) J. Ravichandran, J. Kardel, M. Scullin, J.-H. Bahk, H. Heijmerikx, J. Bowers, and A. Majumdar, Rev. Sci. Instrum. 82, 015108 (2011).
- Fu et al. (2017) Q. Fu, Y. Xiong, W. Zhang, and D. Xu, Rev. Sci. Instrum. 88, 095111 (2017).
- Kedia, Singh, and Chaudhary (2018) S. K. Kedia, A. Singh, and S. Chaudhary, Measurement 117, 49 (2018).
- Yazdani, Kim, and Pettes (2018) S. Yazdani, H.-Y. Kim, and M. T. Pettes, Rev. Sci. Instrum. 89, 084903 (2018).
- Elliott et al. (2007) C. Elliott, V. Vijayakumar, W. Zink, and R. Hansen, J. Asso. Lab. Auto. 12, 17 (2007).
- Ramu et al. (2012) A. T. Ramu, P. Mages, C. Zhang, J. T. Imamura, and J. E. Bowers, Rev. Sci. Instrum. 83, 093905 (2012).
- Burkov et al. (2001) A. Burkov, A. Heinrich, P. Konstantinov, T. Nakama, and K. Yagasaki, Meas. Sci. Technol. 12, 264 (2001).