Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 110 (2017) 281 – 285

1st International Conference on Energy and Power, ICEP2016, 14-16 December 2016, RMIT
University, Melbourne, Australia

Development of flexible thermoelectric cells and performance

investigation of thermoelectric materials for power generation
Pritesh Gokhalea, Bavin Loganathanb, James Crowea, Ashwin Datea, Abhijit Dateb,*
a
Faculty of Mechanical Engineering, Technische Universität Chemnitz, Germany
b
School of Engineering, RMIT University, Melbourne 3000, Australia

Abstract

This paper presents experimental performance results for custom made thermoelectric generator. Two fabrication methods,
powder compaction and painting over flexible substrate have been proposed and investigated. Experimental performance results
have been presented for thermoelectric cells made from alloying Bismuth (Bi), Tellurium (Te) and Antimony (Sb). Powdered P-
type (Bi0.4Sb1.6Te3) and N-type (Bi2Te3) thermoelectric materials were chosen for the construction of the thermoelectric cell.
Experimental results showed that P-type Bi0.4Sb1.6Te3 had Seebeck coefficient of 211.77 μV/°C and while N-type Bi2Te3 had
a Seebeck coefficient of 109.09 μV/°C. The flexible thermoelectric generator has shown linear increase in open circuit voltage
with increase in temperature difference across the cold and hot side.
© 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2017 The Authors. Published by Elsevier Ltd.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 1st International Conference on Energy and Power.
Peer-review under responsibility of the organizing committee of the 1st International Conference on Energy and Power.
Keywords: Flexible Thermoelectric cells; Seebeck coefficient; Semiconductor; Figure-of-merit; thermoelectric generator; Powder compaction,
printing

1. Introduction

Due to the consistent global rise in energy consumption, along with the ever increasing demand for electricity,
heating, refrigerating, etc., greenhouse gas emissions are on the incline globally [1]. Majority of the primary energy
consumption is in form of thermal energy and is exhausted to the atmosphere as low temperature waste heat.

* Corresponding author. Tel.: +61399250612

E-mail address: abhijit.date@rmit.edu.au

1876-6102 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 1st International Conference on Energy and Power.
doi:10.1016/j.egypro.2017.03.140
282 Pritesh Gokhale et al. / Energy Procedia 110 (2017) 281 – 285

Thermoelectric cells (TEC) directly convert heat into electricity via the Seebeck effect of TE materials and are
considered by many to have potential to utilize this low temperature waste heat for producing power [1]. The
Seebeck coefficient (S) can be obtained by applying a thermal gradient across a sample, measuring the differential
voltage (ΔV) across two ends of that sample and having temperatures TH and TC at the respective ends [2]. This is
under the condition that no current flows through the sample whilst measurements are being taken [3].
In order to measure S, two different methods can be utilized, the integral and differential method [3]. In the
integral method, one end is kept at a constant temperature whereas the other end of the sample is slowly heated.
When following the differential method, the entire sample is heated to a higher temperature, with temperature
differentials being created between the ends. The linear voltage drop (ΔV) versus the temperature difference (ΔT)
gives the S for a specific temperature step [3]. This research paper follows the differential method, creating a testing
environment in which the whole sample is slowly heated with ΔV measurements being taken with corresponding ΔT
values, to estimate S. Materials which are thermal insulators and good electrical conductors with high S generally
reflect good TE properties [4]. The laws of physics make this extremely hard to achieve. The Wiedemann-Franz law
states that the “electronic part of thermal conductivity to be proportional to electrical conductivity” and the
Pisarenko relation limits the simultaneous enlargement of Seebeck coefficient and electrical conductivity” [5].
The most common material in today’s TEG is Bi2Te3 (Bismuth Telluride). Over the past 30 years, various alloys
idealized around bismuth telluride have been researched and optimized for TE applications [6]. Recent studies have
shown that high performing p-type Bi0.4Sb1.6Te3 materials have been manufactured via mechanical alloying (MA)
and hot extrusion in the temperature range of 360°C – 450°C. Figure of Merit value of 1.2 at room temperature for
extrusion temperatures of 400°C have been recorded [7].
Power generation through the deployment of TEGs has already been utilised in many areas, including; transport
tools, aerospace facilities and industry utilities [8], where the main focus is utilising waste heat to generate electrical
power. Recent developments have found effective use in harvesting waste heat from the human body for
wristwatches, designed by Seiko and Citizen [9]. There are many desirable features in the deployment of TEGs.
They are extremely environmental friendly in the electricity production for using waste heat as an input source and
they also allow the use of energy efficiently [10, 11]. TE products can be used with a reliable life of 25 years [9],
requiring no maintenance. They are quiet when operating and have no moving parts [12].
Thermoelectrics is an area with a wide scope for further research and experimentation. This research paper looks
into two cost effective fabrication methods, firstly through the exploration of metallurgy powder compaction using a
steel cylindrical template and hydraulic press and second by using painting on flexible substrate. P-type
Bi0.4Sb1.6Te3 and n-type Bi2Te3 alloys are compared and tested using S-type testing rig. Custom made TE
generator is constructed and its performance tested. Results and important findings are discussed throughout this
paper.

2. Thermoelectric materials fabrication

In order to optimize thermoelectric properties, it is essential to calculate and measure weight ratios of the
elements accurately. In order to do this, we must know the molar mass (M) of the TE material as well as the atomic
mass (ma) of the individual elements that make up the TE material. In order to calculate the correct weight
percentage, we divide the total mass of each element by the overall molar mass of the TE material. For simplicity
purposes, 10g powders for both p-type and n-type TE materials were produced by crushing pure bismuth, tellurium
and antimony (approximately 99.9% purity).
Method 1: Accurately measured portions of Bismuth, Tellurium and Antimony are then placed into a glass tube
which is then connected to a vacuum pump to extract any air. While under vacuum the glass tube is heat sealed. The
vacuumed tube is then placed upright in a furnace to melt and form TE alloy. To ensure that the individual elements
are evenly infused, the glass tube is shaken for a few seconds every 20 minutes. The process takes about an hour.
After which, the glass tube is removed out from the furnace and allowed to be cooled to the ambient temperature.
Then the alloy is grinded into fine powder.
Method 2: After determining the required mass of each element required to create 10 grams of powder samples,
the following steps are followed. Firstly, pure elements are weighed out using laboratory scales (accuracy to 0.01g)
to their required mass. Once the desired mass is achieved, the element is then grinded into a fine powder. Now all
 Pritesh Gokhale et al. / Energy Procedia 110 (2017) 281 – 285 283

powdered components are mixed thoroughly to form either p or n type powder. The powder made with either
method 1 or 2 is used to make p and n type legs either by mechanical compaction method or printing on flexible
substrate method.

2.1 TEG Fabrication

Figure 1 shows the layout and the actual flexible thermoelectric generator. 10 grams of P and N type material
powder was individually mixed with 5ml of methanol to form a thick paste. With help of paint brush this paste of P
and N type material was painted on electrically insulating substrate (50 micron paper) to form the TE legs. The p and
n junctions are formed by electrically connecting the ends of p and n type legs using self-adhesive aluminum tape.
Finally polyimide tape is used to seal the pairs of p and n type TE legs and to for a flexible TEG assembly as shown
in the figure 1. The p and n pairs are separated by high temperature flexible electrical insulating material (polyimide
tape).
N-type leg

P-type leg
Electrical
Flexible electrical insulation
Conducting tape
Top and Bottom

Fig 1 Layout of p and n type TE legs painted over flexible substrate and photo of assembled flexible TEG

3. Results and Discussion

P-type TE materials were manufactured and a sample was tested over a ΔT range of 0 – 110 °C. All observed
results were recorded and averages calculated. Figure 2 (a) below shows the voltages produced from TE sample as it
was placed into the testing rig and introduced to a temperature gradient. Figure 2 (b) shows the seebeck coefficient
estimated based on the voltage measurement from figure 2(a). The highest estimated S for sample 1 was 118.33
μV/°C with ΔT of 96.5°C, but it should be noted that S for P-type material is almost constant over the tested range
of temperatures.
14000
140
Seebeck Coefficient (µV/(°C)

12000 Sample 1: Bi0.4Sb1.6Te3

120
10000
Micro Volts (µV)

100
8000
80
Sample 1: Bi0.4Sb1.6Te3
6000 60
4000 40
2000 20

0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Temperature Difference (°C) Temperature Difference (°C)

Fig 2: (a) P-type Temperature difference v/s voltage (b) Seebeck coefficient v/s temperature difference

N-type Bi2Te3 material was tested over a ΔT range of 0 – 160 °C. Figure 3 (a) below shows the open circuit
voltages produced by N-type TE sample as it was placed into the testing rig and introduced to a temperature
gradient. Figure 3 (b) shows the estimated Seebeck coefficient of the N-type material and the highest recorded
Seebeck coefficient for this material was 109.09 µV/°C at ΔT of 36.7°C, but it should be noted that S for N-type
material is almost constant over the tested range of temperatures. Average values of S were estimated to be 110
µV/°C for P-type material and 107 µV/°C for N-type material.
284 Pritesh Gokhale et al. / Energy Procedia 110 (2017) 281 – 285

18000
140
16000

Seebeck Coefficient (µV/(°C)

Bi2Te3 N Type
120
14000
Micro Volts (µV)

12000 100
10000 80
8000 Bi2Te3
60
6000
40
4000
2000 20
0 0
0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160
Temperature Difference (°C) Temperature Difference (°C)

Fig 3: (a) N-type Temperature difference v/s voltage (b) Seebeck coefficient v/s temperature difference

The custom made flexible TEC with 5 pairs of P-type (Bi0.4Sb1.6Te3) TE material and N-type (Bi2Te3) TE
material was fabricated as shown in Figure 1 and experimentally tested for maximum hot side temperature of 75°C
and a constant cold side temperature of 20°C. Test results are shown in Figure 4
40

35
Open circuit voltage (mV)

30

25

20

15

10

0
0 10 20 30 40 50 60 70 80
Temperature difference (°C)

Fig 4: Flexible Thermoelectric Generator – Open circuit voltage versus applied temperature difference

The graph above shows the voltages produced from the custom made flexible TE generator, it can be noted that the
open circuit voltage is constant. But it was observed that bending of the flexible TEG was affecting the open circuit
voltage. The open circuit voltage was over 53% less than expected for the applied temperature difference. This
could be due to,

x Weak contact between conductor and the TE legs. It is highly likely that when observed on a microscopic
scale, there are gaps in contact between the aluminium and TE legs. As a result, performance observed
would be lower than that of an arrangement with perfect contact;
x Painting of TE legs results in porous leg that reduces electrical continuity.

4. Conclusions

Individual legs of TE material were fabricated using powder compaction method and the experimentally
measured Seebeck coefficient for individual p-type leg was 110 µV/°C and for individual n-type leg was 107 µV/°C.
Flexible TE generator was fabricated with a relatively simple low cost painting method. The flexible TE generator
was made by painting 5 pairs of P and N legs on flexible substrate. Ideally it was expected that this 5 pair assembly
would produce an open circuit voltage of around 54mV at temperature difference of 50qC, while the experiments
 Pritesh Gokhale et al. / Energy Procedia 110 (2017) 281 – 285 285

showed that the open circuit voltage was around 25mV at temperature difference of 50qC. The drop in the
performance is due high porosity of the painted legs as compared to the compacted legs used for individual Seebeck
coefficient test. Further the poor connection between the conducting tape and TE legs has also added to the
uncertainty of electrical contact resistance and hence could lead to unpredictable TEG performance.

References

[1] Zheng, X.F., C.X. Liu, Y.Y. Yan, and Q. Wang. A review of thermoelectrics research – Recent developments and potentials for sustainable
and renewable energy applications. Renewable and Sustainable Energy Reviews 2014; 32: p. 486-503.
[2] Zhou, Z. and C. Uher. Apparatus for Seebeck coefficient and electrical resistivity measurements of bulk thermoelectric materials at high
temperature. Review of Scientific Instruments 2005; 762.
[3] Mishra, A., S. Bhattacharjee, and S. Anwar. Simple apparatus to measure Seebeck coefficient up to 900 K. Measurement 2015; 68: p. 295-
301.
[4] Amatya, R. and R.J. Ram. Trend for thermoelectric materials and their earth abundance. Electronic Materials 2012; 416: p. 1011-1019.
[5] Zhang, X. and L.-D. Zhao. Thermoelectric materials: Energy conversion between heat and electricity. Journal of Materiomics 2015; 1: p. 92-
105.
[6] Lee, P.-Y., J. Hao, T.-Y. Chao, J.-Y. Huang, H.-L. Hsieh, and H.-C. Hsu. Thermoelectric properties of nano/microstructured p-type
Bi0.4Sb1.6Te3 powders fabricated by mechanical alloying and vacuum hot pressing. Electronic Materials 2014; 436: p. 1718-1725.
[7] Nagami, Y., K. Matsuoka, T. Akao, T. Onda, T. Hayashi, and Z.-C. Chen. Preparation and characterization of Bi0.4Sb1.6Te3 bulk
thermoelectric materials. Journal of Electronic Materials 2014; 436: p. 2262-2268.
[8] Zheng, X.F., Y.Y. Yan, and K. Simpson. A potential candidate for the sustainable and reliable domestic energy generation - Thermoelectric
cogeneration system. Applied Thermal Engineering 2013; 53: p. 305-311.
[9] Khan, A.A., I. Khan, I. Ahmad, and Z. Ali. Thermoelectric studies of IV–VI semiconductors for renewable energy resources. Materials
Science in Semiconductor Processing 2016; 48: p. 85-94.
[10] Ahıska, R. and H. Mamur. Design and implementation of a new portable thermoelectric generator for low geothermal temperatures. IET
Renewable Power Generation 2013; 7: p. 700-706.
[11] Hsu, C., D. Yao, K. Ye, and B. Yu. Renewable energy of waste heat recovery system for automobiles. Journal of Renewable Sustainable
Energy 2010; 2: p. 105-116.
[12] Tsai, H. and J. Lin. Model building and simulation of thermoelectric module using Matlab/Simulink. Journal of Electronic Materials
2009; 39: p. 2105-2111.