저작권법에따른이용자의권리는위의내용에의하여영향을받지않습니다.

Transcription

1 저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할수없습니다. 변경금지. 귀하는이저작물을개작, 변형또는가공할수없습니다. 귀하는, 이저작물의재이용이나배포의경우, 이저작물에적용된이용허락조건을명확하게나타내어야합니다. 저작권자로부터별도의허가를받으면이러한조건들은적용되지않습니다. 저작권법에따른이용자의권리는위의내용에의하여영향을받지않습니다. 이것은이용허락규약 (Legal Code) 을이해하기쉽게요약한것입니다. Disclaimer

2 Master's Thesis Wearable thermoelectric generator based on printed BiTe legs with a self-heating solar absorber Yeon Soo Jung Department of Materials Science and Engineering Graduate School of UNIST 2018

3 Wearable thermoelectric generator based on printed BiTe legs with a self-heating solar absorber Yeon Soo Jung Department of Materials Science and Engineering Graduate School of UNIST

4 Wearable thermoelectric generator based on printed BiTe legs with a self-heating solar absorber A thesis submitted to the Graduate School of UNIST in partial fulfillment of the requirements for the degree of Master of Science Yeon Soo Jung Approved by Advisor Kyoung Jin Choi

5 Wearable thermoelectric generator based on printed BiTe legs with a self-heating solar absorber Yeon Soo Jung This certifies that the thesis of Yeon Soo Jung is approved. 12/13/2017

6

7 Abstract Wearable thermoelectric generator based on printed BiTe legs with a self-heating solar absorber Yeon Soo Jung School of Materials Science and Engineering The Graduate School Ulsan National Institute of Science and Technology A thermoelectric element is a material and technology that best meets the demands of energy saving in devices by converting heat energy into electrical energy. Its applications include power generation in aviation, semiconductor, biotechnology, consumer electronics industries. Therefore, researches for improving thermoelectric materials and efficiency are actively underway. Meanwhile, the development of wearable electronic devices has generated significant volume of interest in wearable and selfsustainable technology. Self-sustainable technology produces energy from human surroundings and human motion. Such self-sustainable technologies include piezoelectric, triboelectric, and thermoelectric generators (TEGs). Recently, interest in wearable thermoelectric generator (WTEG) has increased with the emerging demand for self-powered mobile electronics and wireless sensor. Various studies of WTEG using body heat and application process have been performed. Conventional WTEGs produce electricity using a difference of temperature between the surface of the human skin and the surrounding temperature, which only focus on the enhancement of the material properties and the flexibility. However, there are low conversion efficiency due to low temperature differences between the body temperature and the atmosphere. Furthermore, the WTEG applied to the human bodies has limitations from practical point of view such as inadequate contact that is usually fixed within 32 C, and irregular temperature with different parts of human body, which result in the inconsistent output power. Another potential inconvenience of this device stems from the fact that it has to be exposed to ambient air to maximize the temperature gradient between the external temperature and the human skin because the efficiency I

8 of the device sharply decreases when there is a temperature equilibrium. Therefore, the temperature difference in WTEG becomes as important as the TE material properties. An ultimate way to increase the temperature difference to improve TE performance is to incorporate solar energy into TEG by adapting solar absorber structure which is called a solar thermoelectric generator (STEG). In terms of the maximum power of the TEG, it is an attractive approach to focus on increasing the ΔT, given that the maximum power is equivalent to the square of the temperature difference by the equation of P max = S 2 (ΔT) 2 / 4R. STEG which integrates a nanocone or pyramid structure on the hot side of a TEG. However for STEG, device and system design complexity has long been identified as a major obstacle to their development, requiring three dimensional and complicated structures such as a vacuum encapsulation, a heat sink structure, and a condenser for realizing the STEG through high thermal concentration for enhancing their efficiency. So, it is not suitable as a wearable device that requires a flexibility and planner structure. Herein, we report a wearable and solar thermoelectric generator (W-STEG) that is formed by a selftemperature gradient through a locally deposited solar absorber consisted of Ti/MgF 2 multilayer structure, through a shadow mask on a flexible in-plane substrate and self-heating under the light. The present W-STEG, integrated wearable with solar thermoelectric, has succeeded in increasing the output power by high temperature difference corresponding to 20 K in actual application with enough wearability. The other issue of the apparent heat transfer losses, including thermal convection and radiation losses, can be eliminated by convection loss through proper insulation, so the temperature difference is more likely to be maximized. II

9 Contents Chapter 1. Introduction of the Basic Working Principle and the Mechanism Behind Thermoelectric Generator Basic principle of thermoelectricity Thermoelectric effect Characterization of thermoelectric generator The strategies to enhance thermoelectric performance Characteristic of Bismuth-Telluride material Various types of thermoelectric generator Conventional thermoelectric generator Printing-based thermoelectric generator Flexible/wearable thermoelectric generator Solar absorber-based solar thermoelectric generator Chapter 2. Wearable solar thermoelectric based on Bi-Te based ink printing technique Experimental methods Synthesis of Bi 2Te 3-based thermoelectric ink Dispenser printing process of the thermoelectric leg Characterization of the thermoelectric generation Simulation of the structure of Ti/MgF 2 superlattice Simulation of the heat flux on the substrate with Ti/MgF 2 superlattice Characterize of thermoelectric ink Viscosity analysis of thermoelectric ink for printing SEM and XRD analysis for Sb2Te4-solder effect on thermoelectric ink Thermoelectric properties of the dispenser printed leg Ti/MgF 2 multilayer thin film based solar absorber Structure optimization of Ti/MgF2-superlattice based on the FDTD simulation SEM and optical analysis of Ti/MgF2-superlattice Heat flux modeling Ti/MgF 2 superlattice dimension optimization based on heat flux simulation Fabrication of wearable solar thermoelectric generator Device structure optimization Thermal imaging of the W-STEG III

10 2.4.3 The characterization of W-STEG Application of wearable solar thermoelectric generator The process of thermoelectric painting Flexibility of the W-STEG Application of wearable solar thermoelectric generator The optimum temperature difference considering the heat flux Conclusion IV

11 List of Figures Figure 1.1. Basic principle of Seebeck and Peltier effect in TE materials. Figure 1.2. Working principle of TE module based on Seebeck and Peltier effect. Figure 1.3. Interrelation of the TE properties; Variation of the transport coefficients as a function of the carrier concentration. Figure 1.4. Increasing the power factor-low dimensionality Figure 1.5. Increasing the power factor by energy filtering effect. Figure 1.6. Crystal structure of bismuth telluride. Figure 1.7. Schematic view of a conventional TEG showing the p / n-type TE legs that are electrically connected in series and thermally arranged in parallel, with interconnecting metal, sandwiched between two ceramic plates. Figure 1.8. Schematic illustration of TEG fabrication via various printing process. (a) Inkjet printing, (b) dispenser printing, (c) screen printing, and (d) stereolithography. Figure 1.9. Schematic illustration of TE device fabrication via dispenser printing on polyimide film and real images of printed TE device. Figure Schematic illustration of all dispenser printed TEG. Figure Schematic illustration of fabrication process via screen printing. Figure (a) Image of the planar TEG for the flexible thermoelectric films and (b) schematic illustration of wearable TEG by Bae et al. Figure A silk-based TEG and its output voltage in real-world application. Figure Illustration of the concept of solar thermoelectric power conversion and conceptual block diagram showing a S-TEG with logic to compute model components, corresponding input parameter and heat flow and the power output of the S-TEG. Figure Schematic of a S-TEG cell structure consisting of a pair of thermoelectric elements with a flat-panel selective absorber that also acts as a thermal concentrator and two bottom electrodes that serve as heat spreaders and radiation shields, and glass enclosure maintaining an evacuated environment. Figure Power factor of various thermoelectric materials Figure Schematic illustration of TE device fabrication via printing process Figure Photograph of Bi 2Te 3 and Sb 2Te 3 columns on a glass fabric of 4 X 4 cm Figure 2.1. Producing a TE ink based on an inorganic soldering process Figure 2.2. Photograph of dispenser printer Figure 2.3. (a) Schematic of TE power measurement set-up. (b) Temperature on the hot and cold side corresponding to T indicating temperature reliability. V

12 Figure 2.4. Rheology of the Bismuth Telluride based TE paste. Figure 2.5. SEM images of the TE materials (a) without and (b) with the Sb 2Te 3-based inorganic solder after heat treatment. Figure 2.6. XRD patterns of (a) BST and (b) BTS sample with different sintering temperatures Figure 2.7. The cross-sectional area of thermoelectric leg by scanning the thermoelectric leg sample with alpha-step Figure 2.8. A linear fit of the resistance as a function of the distance between TE leg patterns by transmission line method. Figure 2.9. The relationship between the voltage and the T. Figure (a) IV curves at different ΔT. (b) Voltage and power as a function of ΔT. Figure (a) Simulated results of the amount of heat by the solar absorber which depends on Ti / MgF 2 thickness. (b) Absorbed energy according to the period in which the optimum thickness is applied. Figure (a) A cross-sectional SEM image of Ti/MgF 2 solar absorber. Figure Optical properties of (absorptance[a], reflectance[r], transmittance[t]) of solar absorber[sa]. Figure Calculation of heat capacity of PI with Ti/MgF 2 superlattice deposited. Figure Schematic diagram of heat transfer model Figure Schematic illustration of the heat flow through the interface. Figure (a) Simulated results of temperature gradient according to solar absorber width. (b) Temperature measurement using thermocouple every 2mm intervals according to solar absorber width. The inset shows the measurement schematic Figure Substrate temperature measurement with retention time Figure Schematic diagram of the manufacturing process for the W-STEG and a process for (a) depositing a Ti/ MgF 2 multilayer solar absorber on flexible PI substrate, (b) dispenser printing of thermoelectric ink on PI substrate with a solar absorber and (c) measuring W- STEG power under AM 1.5G condition. Figure SEM images of n-type (a) and p-type (p) TE paints sintered at various temperatures Figure Power and voltage comparison according to (a) solar absorber width and (b) single printed leg length. Figure Temperature dependence of calculated lattice thermal conductivities of BST/BTS painted samples using the modified formulation of the effective medium theory Figure (a) Thermographic image of W-STEG (b) with and (c) without light blocking to TE legs Figure Low-magnification SEM images of fractured structure (a) and surface of painted sample. (b) n-type, (c) p-type. The red circles show the micro-scale pores in the samples. VI

13 Figure Stability of the internal resistance of the bending cycle repeatedly along the bending radius Figure The schematic illustration of the stress during bending cycle with / without parylene C. Figure The cross-sectional SEM image of the parylene C coated TE legs, which shows the micro crack filling. Figure Photograph of the W-STEG attached to clothes and windows. Figure Experimental results, voltage and T, of wearable TEGs application. Figure Voltage-time curve showing immediate response to light and retention of the voltage over time. Figure Schematic illuttration of the strategies to enhance temperature difference. VII

14 List of Tables Table 1. The variation of XRD peaks of Bi 0.4Sb 1.6Te 3 and Bi 2Te 2.7Se 0.3 samples. Table 2. The resistance and the electrical conductivity with the length of TE legs. Table 3. The detailed constants of the heat transfer model system. Table 4. The comparison of experimental results of wearable TEGs application. VIII

15 Chapter 1. Introduction of the Basic Working Principle and the Mechanism Behind Thermoelectric Generator 1.1 Basic principle of Thermoelectricity Thermoelectric devices, which are semiconductor systems that can directly convert waste heat into electrical power, are seen to have the potential to make important contributions to reduce CO 2 and greenhouse gas emissions and provide cleaner forms of energy. They are progressing for power generation because they do not require moving parts Thermoelectric Effects The thermoelectric effect represents the interaction and transformation of the heat into electricity, defined by Seebeck and Peltier effect. The Seebeck effect generates thermoelectric power when there is a temperature gradient between the junctions of different kinds of materials, as shown in Figure 1.1. At the atomic scale, when TE material is heated, the charge carriers spread to the cold side which leads to the production of the electrostatic potential energy. The reverse reaction of the Seebeck effect is called the Peltier effect [1]. Based on these two effects above, thermoelectric generators have been utilized in cooling and in power generation. Fig Basic principle of Seebeck and Peltier effect in TE materials. The base module of the thermoelectric device has a thermocouple consisting of two thermocouples based on Seebeck and Peltier effect, as shown in Figure 1.2. When a heat passes through a thermocouple, the current and voltage are generated by the load resistor R L. The Seebeck effect is also proportional to the ratio of generated voltage (V) to temperature gradient (T). 1

16 S = dv dt (1.1) In other words, the Seebeck effect is the voltage that occurs when two materials are in thermal gradient. The output voltage is proportional to the temperature gradient ( T) between the two junctions. V oc = α n p T (1.2), where α n-p is a proportional constant and is called a Seebeck coefficient. The α is a quantity that is relative and is related to the materials properties. Seebeck coefficient unit is V / K, and it is positive when the majority of the charge carriers are holes, and negative when they are electrons. In general, the insulator has a high Seebeck coefficient followed by semiconductors, semimetals and metals. Fig Working principle of TE module based on Seebeck and Peltier effect Characterization of thermoelectric generator The performance of the thermoelectric device directly proportional to the T and dimensionless figure of merit; ZT, which is generally defined by ZT = S2 σ T (1.3) k 2

17 , where (S) is Seebeck coefficient, (σ) is electrical conductivity, (k) is the thermal conductivity and (T) is absolute temperature. The quantity S 2 σ is typically called a power factor (PF). The temperature gradient and the dimension of thermoelectric legs are related to the output power [2]. The output power of a TEG can be defined as: P max = V oc 2 4R in (1.4), where V oc is the voltage of an open circuit and R in is internal resistance of the device, which is given as R in = ρ ml A (1.5), where ρ is the electrical resistivity, L is length of the thermoelectric leg, and A is the cross-sectional area of the leg [3]. Also, it should be noted that P max is equivalent to the temperature difference square from equation (1.2) and (1.4). P max = α 2 n p( T) 2 (1.6) 4R in The output power of the thermoelectric generator is a function of the thermoelectric parameters. It depends on the pairs of thermocouples (n), Seebeck coefficient of the pair of thermocouples S 1(p-type), S 2(n-type), and the temperature difference ΔT across the generator. V output = nδt(s 1 S 2 ) (1.7) Therefore, the electrical resistance of the device must be low to for maximum power output and it requires short length of leg [4]. If the length of leg is too small, it is hard to maintain the temperature gradient in the device. This causes a trade-off between length of thermoelectric leg and the output voltage, and thus the device must be designed to account for changes in output depending on leg length. The thermoelectric efficiency can be expressed as (1.8), where T c is the temperature of the cold side, T h is the temperature of the hot side, and the effective ZT of the thermoelectric material between T c and T h. Increasing efficiency requires large temperature gradient and thermoelectric properties [5] The strategies to enhance thermoelectric performance The challenge to achieve high ZT values is to produce high electric conductivity, Seebeck coefficient and low thermal conductivity, simultaneously. These parameters are determined by the charge carriers and mobility which are not controllable independently. In order to maximize ZT; either the thermal conductivity minimized or the electrical conductivity 3

18 and Seebeck coefficient is maximized. Power factor, S 2 σ, Seebeck coefficient and electrical conductivity is the first step to maximize ZT. A large ZT produces a large voltage and current. To maximize PF, development of materials with new nanoscale electronic structure and doping of the thermoelectric material is crucial. Another way to maximize ZT is to optimize the lattice thermal conductivity. (1.9), where k el is the carrier thermal conductivity. To produce a large temperature gradient, thermal conductivity must be low. Manipulation of thermal conductivity can be achieved by the nanostructuring and solid-solution alloying to produce materials with low thermal conductivity [6]. As shown in Fig.1.3, according to Wiedeman-Franz law given by k el = LσT = neµlt (1.10), high electrical conductivity would increase the electronic contribution. Based on Wiedeman-Franz law, electrical conductivity is directly proportional with the thermal conductivity k el, therefore high electronic conductivity is not always suitable [7,8]. If the carrier concentration is low, it has high thermopower but it has very low electrical conductivity and thermal conductivity similar to that of an insulator and if the carrier concentration is high, it shows considerably low Seebeck coefficient and have high thermal conductivity similar to that of a metal. Fig Interrelation of the TE properties; Variation of the transport coefficients as a function of the carrier concentration. In the synthesis of solid solutions, we can also introduce the concept of point defects. The point defects in solid solutions provide entropy throughout the crystal structure which induces strong scattering of phonons leading to lower thermal conductivity and therefore a large ZT value. The other way to lower the thermal conductivity is using PGEC approach [9], where electron carrier passes 4

19 through the structure and heat carrying phonons are deflected off the structure. The name comes from the structure having a low k lat similar to a glass and simultaneously a high electrical conductivity similar to a crystal. This approach result in a dramatic reduction of lattice thermal conductivity by phonon damping effect. Another way to lower thermal conductivity is to utilize Skutterudites [10], Clathrates [11], Half-Heusler [12] and metal oxide [13]. To increase the power factor, we can approach with low dimensionality. In 1993, Dresselhaus implemented a theoretical analysis on enhanced ZT value for 1D quantum wires and 2D quantum wells [14,15]. Enhanced Seebeck coefficient results from the increase in the density of state (DOS) near fermi level due to quantum confinement effects from low-dimensional materials. Rapidly changing DOS would result in a high thermoelectric power, and sharp changes in DOS suggest that the carriers have flat dispersion behavior (Fig. 1.4). Fig Increasing the power factor-low dimensionality. Another way to increase the PF through the breaking the trade-off relationship between Seebeck and electrical conductivity is to use energy filtering effect, as shown in Fig By applying energy-filtering method, energy barriers deflect the electrons with low energy and thus increasing the average heat transported per carrier. Grain boundary interfaces are more effective when the height of the barrier is uniform in the grain. The relaxation times (τ) can be described as τ τ 0 E r, where the r represents the scattering parameter [16]. An increase of the relaxation times increase the differential slope of the conductivity which further enhances the Seebeck coefficient. 5

20 Fig Increasing the power factor by energy filtering effect Characteristic of Bismuth Telluride material Bismuth telluride has a rhombohedral structure with perpendicularly stacked alongside to the c-axis of the sequence. Layers of -Te (1) -Bi- and -Bi-Te (2) - are attached together by strong ionic-covalent bonds and covalent bonds, while the -Te (1) -Te (1) - layer is held by Van Der Waals bond [17]. The crystal structure of bismuth telluride shown in the fig The mechanical strength of Bi 2Te 3 is very weak because of the existence of gap along the basal plane perpendicular to the c-axis. Thus, the mechanical strength and the transport properties of BiTe-based TE materials have strong anisotropy [18]. In recent years, commercially available TE devices based on bismuth telluride are widely available and are mainly used for local heating and cooling applications. TE devices for large-scale waste heat recovery have recently been commercialized, and feasibility studies have been reported in industrial processes. 6

21 Fig Crystal structure of bismuth telluride. 7

22 1.2 Various types of thermoelectric generator Conventional Thermoelectric generator Thermoelectric generator works based on the Seebeck effect as discussed in Since these devices are electrically connected in series, the voltages generated across n and p-type are integrated together. An increase of n and p type pairs improve the output voltage. Figure 1.7 shows a picture of a conventional thermoelectric generator. Generally, ceramic substrates work as substrates to keep the temperature gradient across thermoelectric elements and to maintain electrical insulation between thermoelectric elements and electrodes. When designing these rigid type TEGs, it is important to know the temperature differences that can be obtained from low waste heat. In other words, a conventional TEG with a rigid ceramic plate unfortunately causes considerable heat loss due to incomplete contact between the surface of heat source and the thermoelectric module in practical applications [19-21]. It should be noted that most of the heat source of the thermoelectric generator has an irregular shape, and planar structured thermoelectric device composed of cubic blocks fails to achieve the desired contact. It is more important to mitigate heat loss than to enhance the performance index of the material. Fig Schematic view of a conventional TEG showing the n- and p-type TE legs which are electrically connected in series and thermally arranged in parallel, with interconnecting metal, sandwiched between two ceramic plates Printing-based Thermoelectric Generator The desire to use energy resources more efficiently has led to the growth of research and development 8

23 of TE materials and devices, particularly polymer-based TE materials that allow for low temperature waste heat recovery with flexible and inexpensive equipment [22]. Polymer-based TE devices are often manufactured through printing method. Printing technology used in the manufacture of thermoelectric devices includes inkjet, screen, dispenser, and 3D printing, stereolithography, etc., as shown in Fig Fig Schematic illustration of TEG fabrication via various printing process. (a) Inkjet printing, (b) dispenser printing, (c) screen printing, and (d) stereolithography. The figure is reproduced with permission from Ref. [22] The Fig. 1.9 shows (a) a fabrication process of TEG by dispenser-printer and picture of (b) dispenserprinted planar thermoelectric generator on a polyimide film and(c) prototype 50-couple device with electrical leads. In this work, a printable thick-film thermoelectric material was successfully synthesized for flat arrays with high aspect ratio TEG fabrication [23,24]. Thermoelectric properties for n- and p- type thermoelectric legs show ZT value of 0.19 and 0.18, respectively. The linear-type of thermoelectric legs were dispenser-printed on a flexible polyimide substrate and the device was rolled forming a round TEG. The power generated 10.5 μw at mv for ΔT of 20 K. The result shows an output power density of 75 μw cm -2. The results show a possibility for developing low-cost self-sustainable energy device. 9

24 Fig Schematic illustration of TE device fabrication via dispenser printing on polyimide film and real images of printed TE device. The figure is reproduced with permission from Ref. [24] The figure 1.10 demonstrates the complete dispenser printing thermoelectric device manufacturing process. 3D thermoelectric legs on a polyimide substrates (Kapton) was fabricated by a dispenser printer [25]. The thermocouples were printed in a three dimensional cube. The Kapton substrate is printed at the same height to physically bolster the top electrode, so a vertical TEG can be 3D-printed. In conclusion, dispenser printing can produce TEGs in a 3D structure. However, the high contact resistance of the device must be optimized. Evaporation and sputtering can provide solutions for depositing conductive electrodes. 10

25 Fig Schematic illustration of all dispenser printed TEG. The figure is reproduced with permission from Ref. [25] The figure 1.11 shows the schematic illustration of overall manufacturing method for the flexible thermoelectric films via screen printing on flexible substrate [26]. The film present maximum ZT of 0.43 and outstanding flexibility with little change in thermoelectric property after bending cycles of 150 times. A technology capable of opening many applications for low-temperature energy harvesting, a flexible thermoelectric device that are printed produces a power density of about 4.1 mw / cm 2 at a relatively large temperature gradient of 60 C. The thermoelectric performance and device may be further improved by optimization of sintering temperature and the formulation of ink. 11

26 Fig Schematic illustration of fabrication method via screen printing. The figure is reproduced with permission from Ref. [26] Flexible/Wearable Thermoelectric Generator Flexible/wearable TEGs have the potential to be used to generate power from waste heat in compact and wearable devices. Flexible/wearable TEGs are also receiving more attention in biomedical applications. The wearable thermoelectric devices proposed by Bae et al. incorporate Te-PEDOT: PSS hybrid composites [27]. This device has a high Seebeck coefficient due to the properties of Te nanorods (Fig. 1.12). However, the electrical conductivity was very low. The slightly improved electrical conductivity due to the H 2SO 4 treatment is believed to increase the number of charges due to the removal of PSS, but still has a low value. Large area applications have been achieved using organic species by a printing H 2SO 4-treated Te PEDOT: PSS composites with a power factor of 284 μw m -1 K - 2. The printed thermoelectric array was demonstrated with a relatively low voltage of 2 mv and an output power of nw in response to heat from the human body. Fig (a) Image of the planar TEG for the flexible thermoelectric films and (b) schematic illustration of the wearable TEG by Bae et al. The figure is reproduced with permission from Ref. [27] 12

27 Z. Lu et al. developed a wearable TEG based on silk fabrics to harvest energy from the human skin [28]. Hydrothermally synthesized n-type Bi 2Te 3 nanotubes and p-type Sb 2Te 3 nanoplates were deposited on silk fabrics (Fig. 1.13). The silk fabric-based TEG consisting of 12 thermocouples had a maximum voltage of ~ 10mV and output of ~ 15nW. The performance of the device did not deteriorate after 100 bending cycles. This research provides a new outlook to the development of fabric-based TEG that can be integrated with rechargeable batteries to fabricate a self-charging system for wearable devices in real-world applications. Fig A silk-based TEG and the output voltage in real-world application. The figure is reproduced with permission from Ref. [28] For performance related to the printed TEG and flexible/wearable TEG reviewed in and 1.2.3, the maximum output includes a wide range of several pw to µw, significantly lower than the power produced by traditionally manufactured inorganic TEG operating at W and kw range. In case of thin film wearable TEG, efforts have been made in various aspects such as material development for flexibility by using glass fabric or silk fabric, however there are low conversion efficiency due to low temperature differences induced by the large thermal resistance that occurs at the human body/teg and TEG/ambient interface which is important in terms of real application presented by Ozturk group [29]. On the other hand, this field can take advantage of standardized characterization techniques to make meaningful comparisons of materials, device geometry, and manufacturing methods. Although the conversion efficiency is much lower than that of inorganic materials, organic and hybrid TE materials are well-suited for printing fabrication techniques for large-scale production of inexpensive, thin and flexible TE devices. These devices can supply power to the wearable electronics and enable waste heat recovery on non-planar surfaces. Due to high efficiency and flexibility it can be utilized in irregular surface of the heat source which is desirable compared to the heavy and brittle nature of inorganic TE 13

28 materials Solar Thermoelectric generator Most of the world's greatest solar resources are focusing to develop technologies that can convert solar energy into electricity in a stable and autonomous way. Solar power was the best way to convert sunlight into electricity. Thermoelectric materials are used in solar thermal applications to utilize in space satellites, automobiles, and recent solar thermoelectric generators (STEGs). Seebeck effects and promising solar-to-power conversion techniques based on high heat concentration allow for a wider range of applications. These STEGs have been focused on the core development of thermoelectric materials and the development of STEG module design. The STEG system has shown great interest in concentrated and non-condensing systems and has been used in hybrid configurations with solar and photovoltaic systems. Figure 1.14 shows general conversion of solar thermoelectric power. Solar-powered thermoelectric cells consist of a possible optical concentrator, a solar absorber, such as a wavelength selective surface that absorbs solar radiation and concentrates heat intensively on n- and p-type thermocouples. The other side of the device is connected to a Cu electrode used to connect the electrical circuit to extract power from the system; W. Zhu et al presented a multifunctional thin film TEG for autonomous light sensing by incorporating a Fresnel lens into the device for energy concentration. With the condensing system, the T can be obtained up to 40 C [30]. Fig Illustration of the conversion of solar thermoelectric power and conceptual block diagram showing a STEG with logic to compute model components. The figure is reproduced with permission from Ref. [30] 14

29 Kraemer et al. developed the solar power generator achieved a maximum efficiency of 4.6% under the condition of AM 1.5G (1000Wm 2 ) [31]. Efficiency is 7-8 times higher than previously reported for flat panel STEG, an innovative design that uses high performance nanostructured thermoelectric materials and spectrally selective solar absorbers to use high thermal concentration (Fig. 1.15). As reviewed above, there is the advantage of producing high T through the STEG theoretical design and various types of STEG, however it is not suitable for wearable TEGs due to the design with 3D structure with bulky components such as a vacuum enclosure, condenser lens, and heat sink. However, solar absorber system used in STEG can provide a new approach to apply with future wearable TEG for improving the efficiency of the device. Fig Schematic of a S-TEG cell structure consisting of a pair of thermoelectric elements with a flat-panel selective absorber. The figure is reproduced with permission from Ref. [31] We have studied the thermoelectric phenomenon and principle based on energy harvesting, and briefly reviewed the history of various TEG systems according to the trend of energy harvesting from conventional TEGs to solar thermoelectric devices and wearable thermoelectric devices. In recent years, flexible TE applications, which is wearable, are emerging as one of the effective energy harvesting method. However, the thermoelectric performance of wearable and flexible TE applications is less than 15

30 that of conventional thermoelectric devices, and high operating efficiency cannot be achieved due to the slight temperature difference in real application. We need a new approach to solve this issue. We need to propose new approaches such as introducing effective heat sources applicable to wearable products and designing new structures. 16

31 Chapter 2. Wearable solar thermoelectric generator based on Bi-Te based ink printing technique 2.1 Experimental methods Synthesis of Bi2Te3-based thermoelectric ink Sb 2Te 3-based inorganic binder was synthesized by mixing Sb and Te ball-milled particles in a cosolvent mixture of ethanethiol and ethylenediamine in a stoichiometric ratio of 2: 8, and was stirred at least 24 hours to dissolve all Sb and Te particles in the solvent [32,33]. Acetonitrile was added to precipitate the solvent mixture and it was centrifuged at 7,500 rpm for 10 minutes. Bi 2Se 0.3Te 2.7 for n- type and Bi 0.4Sb 1.6Te 3 for p-type were produced by a mechanical alloying process through ball milling for 5 hours, and sieved at 45 µm. To produce TE ink, the precipitated Sb 2Te 3 solder and the BST/BTS TE powder by the mechanical alloying process were dispersed in a glycerol at a weight ratio of 1: 4: 4 and was mixed for a half hour with a planetary centrifugal mixer (THINKY) to prevent nozzle clogging owing to chunks. Fig Producing a TE ink based on an inorganic soldering process Dispenser printing process of the thermoelectric leg Thermoelectric legs were printed using home-made equipment of a dispenser printer on the polyimide substrate (Fig. 2.2). The ink was pressed out of the dispenser at a pressure of 0.1 MPa. Dispenserprinting was accomplished by using the LabVIEW program to move the stage in 3-directions x, y, and z with 1 μm spatial resolution. The PI substrate underwent ultraviolet-ozone (UVO) treatment to enhance the adhesion between the TE ink and the substrate. The dispensed TE ink underwent a drying 17

32 process at 90 C to 150 C for 2 hours to dry off excess glycerol without cracking of the surface. Secondly, sintering at 430 C was performed for 2 minutes to improve the density of the material. The morphology of the thermoelectric materials were analyzed by XRD patterns and SEM images. Fig. 2.2 Photograph of dispenser printer Characterization of the thermoelectric generation The Seebeck coefficients (α) of the TE legs were measured depending on the T across the leg by a home-built TE measurement setup as demonstrated in fig I-V characteristics were measured by connecting silver electrodes to the TEG. Four-terminal sensing system was utilized to measure the electrical conductivity and Seebeck coefficient. Using the equation σ = L / (R A), where R is the internal resistance, L is the length of the TE leg, and A is the cross-sectional area of the TE leg, the electrical conductivity σ was calculated. The cross-sectional area of TE leg was estimated by integrating the value obtained by scanning the cross-section of the TE leg with an alpha-step. Fig (a) Schematic of TE power measurement set-up. (b) Temperature on the hot and cold side corresponding to T indicating temperature reliability. 18

33 2.1.4 Simulation of the structure of Ti/MgF 2 multilayer thin film The solar absorber is a Ti / MgF 2 multilayer thin film, optimized by transfer matrix method [34]. In order to optimized Ti / MgF 2 multilayer, the total energy absorbed was calculated as a function of the thickness of Ti / MgF 2 superlattice. Also, the stack cycles in the Ti / MgF 2 superlattice was used to calculate the amount of heat absorbed. The optical properties of Ti / MgF 2 multilayer thin films were calculated based on the Maxwell's equation with appropriate boundary condition and refractive index [35,36] Simulation of the heat flux on the substrate with Ti/MgF 2 superlattice A heat transfer model was used for the calculation of temperature distribution when solar radiation was irradiated on a PI substrate with optimized solar absorbers and simulation study was performed using COMSOL software. The heat transfer of the interface is used to calculate temperature and radiation intensity fields. Heat transfer interfaces and thermal multi-physics couplings can be used for conduction, convection or radiation as well as thermal transfer modeling by complex heat transfer. The heat transfer model was designed, which including conduction, convection and radiation. The temperature equations defined in the solid region correspond to the differential form of the Fourier's rule, which may include additional contributions such as heat sources. The devices analyzed in this study are thermal conduction in thin structures and heat transfer, and heat transfer models at the thin shells interface are used. The thin conductive layer model is enabled by default at all boundaries. It is available to use any function that includes other boundary contributions, such as inter-surface radiation. The temperature equation defined in the shell corresponds to the form of a tangential derivative of Fourier's law that may include additional contributions such as heat sources. The heat transfer model simulation was performed on a PI substrate with a thickness of 20 μm, a width of 40 mm and a length of 40 mm. The thickness of the solar absorber was 0.5 µm, the width was 5 to 20 mm, and the length was 40 mm. Also, we assumed that sunlight hit the surface of the substrate and the absorbed energy is converted to heat. 2.2 Characterization of thermoelectric ink Viscosity analysis of thermoelectric ink for printing With dispenser printing, flexible patterns can be printed using a TE ink. It is important to study the rheological behavior of the TE ink since the viscoelasticity of the ink has a significant effect on the 19

34 shape retention and rigidity of the printed pattern [37-39]. If the viscosity of the TE ink is higher than certain limit, the nozzle is clogged since the ink is not ejected at an appropriate pressure. However, with a low viscosity of the TE ink, which will not be dispensed by the pressure exerted by the distributor, but will begin to drop due to gravity. Figure 2.4 shows rheology behavior and the viscosity of the BST/BTS TE ink along to applied shear rate. The equation of the shear rate (γ ) applied to the TE ink discharging through the circular nozzle is given by [40]: γ = 4Q πr 3 (2.1), where Q is the flow rate of the TE ink and r is the radius of the circular nozzle. When the TE ink was dispenser printed for Q is cc / s and R is 115 μm, the shear rate was calculated as 117 s -1. The resulting ink has a viscosity of 1730 cp, and it was suitable for dispenser printing [38]. Fig Rheology of the TE ink along to applied shear rate SEM and XRD analysis for Sb 2Te 4-solder effect on thermoelectric ink The BST/BTS ink printed on the PI substrate underwent two heat treatment processes. First, the drying glycerol solvent of the ink is carried out at 90 C to 150 C for 2 hours. Secondly, the dried ink was sintered at 430 C for 1-2 minutes to improve the rigidity and the density of the final product. Figure 2.5 shows SEM images of sintered TE legs with and without Sb 2Te 3 binder. The Sb 2Te 3 binder in this work act as a sintering aid, further improving the electrical conductivity of the TE material. The excess Te in the inorganic binder is melted and becomes a liquid phase at high temperatures which fills in the pores and voids of the particles. 20

35 Fig SEM images of the n-type (BTS) and p-type (BST) dispenser printed TE materials (a) without and (b) with the Sb 2Te 3-based sintering additive after sintering at 450 C. Figures 2.6 show the XRD patterns of the Bi 0.4Sb 1.6Te 3 and Bi 2.0Te 2.7Se 0.3 samples with different annealing temperature from 200 C to 450 C. XRD peaks of each temperature correspond with the previously reported peaks of the single crystal sample [41,42]. 21

36 Fig XRD patterns of printed (a) p-type and (b) n-type sample with different sintering temperatures. As shown in Table 1, the FWHM of the main diffraction peak (015) decreased monotonically with increasing sintering temperature. Particularly, the sample annealed at 450 C exhibits a very low FWHM of 0.13, suggesting that the sample was completely densified by heat treatment. As described above, the excess Te becomes liquid phase and the crystallinity of the sample is significantly improved due to the liquid phase sintering [43,44], which corresponds well. That is, Sb 2Te 3-based solder is fully integrated into the host because of its good composition with the host. Table 1. The variation of FWHM value of (015) XRD peaks of p-type Bi 0.4Sb 1.6Te 3 and n-type Bi 2Se 0.3Te 2 samples Thermoelectric properties of the dispenser printed leg The thermoelectric properties of TE legs fabricated under optimum rheology and dispensing condition. To measure electrical conductivity, silver paste was painted on the both end of the TE legs. The crosssectional area was about 0.06 mm 2, which was calculated by alpha step; the width of the printed leg was 1.5 mm, the average height was about 40 μm as shown in Fig

37 Fig The cross-sectional area of TE leg by scanning the TE leg sample with alpha-step We used a similar method to the traditional transmission line method for accurate conductivity measurement. The contact resistance was measured as a function of length change from 8 to 38mm. Fig Linear fit of the resistance with different length of leg in mm As demonstrated in Table 2, the BST/BTS TE legs have electrical conductivities of S / m. Also, the contact resistance between the TE leg and the silver contact was estimated about 0.22 at room temperature. 23

38 Fig. 2.9 demonstrates the TE voltage obtained at a temperature difference of 0 to 10 C. The Seebeck coefficients of the p- and n-type TE legs are and μv, respectively. Fig Dependence of the voltage on T for p / n-type TE legs. I also measured power characteristics of 10 pairs of printed TE legs. The contact resistance was considered in the measurement setup. When the TE leg was directly connected to a silver electrode, the total contact resistance was decreased about 5%. Figure 2.10 shows the IV curve obtained by varying T from 5T to 30K. The schematic illustration of the measurement setup was demonstrated in the inset of the figure. The maximum output of the TEG was measured when the load resistance was at around 140 Ω. Figure 3d shows the maximum voltage and power at the matched load resistance measured according to the T value of the TE generator. The output voltage increases linearly with T. 24

39 Fig (a) Thermoelectric IV curves of 10 pairs of TE legs at various ΔT. (b) Output power and voltage as a function of temperature difference. 2.3 Ti/MgF2 multilayer thin film based solar absorber The metal / dielectric multilayer film structure has been widely utilized as the solar absorber [45-47]. In principle, the metal layer thickness should be less than one skin depth. In other words, it should be thin as several nanometers to allow for a partial transparency. By inserting a dielectric material in the middle layer, effectively reduce the reflectance and induce the absorption by interference effects, which can create an effective absorber. Thus, the thickness of each layer should be optimized by simulation considering the constructive interference of each layer Structure optimization of Ti/MgF2 multilayer film based on the FDTD simulation In this study, solar absorber structure of metal / dielectric multilayer was introduced to enhance thermal gradient. This structure offers a high flexibility and solar absorption, and thus we implemented this structure with MgF 2 as the dielectric and Ti as the metal. Optimum structure of periodic Ti / MgF 2 multilayer film is needed to absorb sunlight up to maximum. And the structure was calculated by considering the solar irradiance spectrum. By multiplying the solar irradiance spectrum to the calculated absorption spectrum for a given multilayer film, we obtained the solar absorption spectrum [48]. We calculated the area ( Λ(λ)dλ) of the solar absorption spectrum according to the thickness of each layer, Ti and MgF 2. As demonstrated in Fig. S4a, the highest total solar absorption was observed at a thickness of 7.3nm for Ti and 96.5 nm for MgF 2. As the number of stack period increases, the number of metal layer that involved in an absorption also inceases, and then the absorption performance of the solar absorber is saturated near the ideal 1 sun absorption (1000 W/m 2 ). Since there is no significant difference in the performace of the solar absorber over the stack number value of five, five periods were 25

40 chosen to minimize the thickness. As a result, the optimal structure consisting of the thickness of 7.3 nm Ti and 96.5 nm MgF 2 of five-period layers was obtained, and it was proved to be enough period to absorb 95% of light (Fig. 2.11). Fig (a) Simulated results of the amount of heat absorbed by the solar absorber which depends on the Ti / MgF 2 thickness. (b) Absorbed energy according to the period in which the optimum thickness is applied SEM and optical analysis of Ti/MgF2-superlattice Figure 2.12 shows a SEM image obtained from cross section of the Ti/MgF 2 multilayer made of 5- period of optimal thickness by repetitive cross deposition. The Ti/MgF 2 multilayer film exhibited uniformly and smoothly deposited structure. The interfaces between the Ti and MgF 2 layers were well defined. As seen in Fig. 2.12, each bright and dark deposited layers correspond to Ti and MgF 2 layers, respectively. The total depositing thickness for the Ti/MgF 2 multilayer film was about 500 nm. Fig A cross sectional SEM image obtained from Ti/MgF 2 multilayer. 26

41 Figure 2.13 shows the optical properties of the Ti/MgF 2 multilayer film measured from the visible to the infrared region. Though the film was only about 500 nm in thickness, high absorbance spectra of ~ 98% can be observed near 550nm wavelength of visible region with high solar intensity and absorbance more than 90% in the region of 500 nm to 1000 nm. The transmitted light was as low as 1% and reflected light is less than 20%. At the wavelength of 520 nm, the sunlight had the highest intensity with low reflectance at 2.2% and the absorbance measured above 97.5%. It suggests that the high absorption energy can be obtained by having a high absorption spectrum in a wavelength band with considerable intensity in a solar spectrum, which means efficiently absorb the sunlight. Fig Optical properties of (absorbance[a], reflectance[r], transmittance[t]) of solar absorber[sa]. Using the experimental result of absorption spectra, the absorbed energy of Ti / MgF 2 and PI films directly to heat is W/m 2 and W/m 2, respectively (Fig.2.14). To obtain the total solar absorption spectrum, the solar irradiance was multiplied to the measured absorbance spectrum. The area of the solar absorption spectrum reflects the total solar absorption which directly relates the heat generation of the solar absorber. 27

42 Fig Calculation of heat capacity of PI with Ti/MgF 2 superlattice deposited 28

43 2.3.3 Heat flux modeling The schematic illustration of the heat transfer mechanism is shown in Fig We calculated the three-dimensional temperature distribution of the device for the steady state solution by performing the simulation based on heat transfer analysis. The solar absorber and PI substrate lengths were 4 cm 2 cm and 4cm 4cm 2, respectively, and the total input of heat to the substrate was 1.6 W, equivalent to the output power of the steady state in energy conservation law. As demonstrated in Fig. 2.16, out of 100% of the sunlight irradiated to the absorber, only 95.6% is absorbed to the absorber and only 22% is absorbed to the PI substrate. The heat is transferred from the solar absorber which is the hot side, to the PI substrate. In this case, energy losses from PI substrate and solar absorber due to convection and radiation are 72.6% and 42.4%, respectively. Fig Schematic diagram of heat transfer model Fig Schematic diagram of the heat flow through the interface based on heat transfer model. I = Incoming solar energy, R = reflected solar energy, T = transmitted solar energy, Q = heat Transferred, conv = Convection heat transfer, rad = radiation heat transfer. 29

44 2.3.4 Ti/MgF 2-superlattice dimension optimization based on heat flux simulation Since the absorber is activated at high temperature gradient between two substrates, the temperature distribution of the two substrates is very important. As demonstrated in Fig. 2.17, the simulation and experimental results shows the temperature distribution with the solar absorber width. Table 3 shows detailed material parameters of the model system. The solar absorbers obtained from FDTD simulations of structural optimization are more reliable and operational. Steep temperature gradients were observed at the two substrates, and ΔT between these two substrates increased with the increase of absorber width. The abrupt temperature difference at the edge is because the thermal resistance of the PI is much higher than that of the solar absorber, therefore increasing the thermal resistance of the PI, which promotes a rapid temperature change. Furthermore, the heat absorbed by the solar absorber increases with the increasing solar absorber width, however the heat fluxes of the convection and radiant become smaller. Fig (a) Simulated results of temperature gradient according to solar absorber width. (b) Temperature measurement using thermocouple every 2mm intervals according to solar absorber width. The inset shows the measurement schematic. 30

45 Table 3. Detailed parameters of the model system [49-54]. Fig Substrate temperature measurement with retention time 2.4 Fabrication of wearable solar thermoelectric generator BST/BTS TE legs were printed on a flexible substrate which was locally deposited by photovoltaic layer as shown in fig T was generated by the hot side of the local solar absorber and cold side of the PI substrate. The output power characteristics were optimized for the solar absorber width and the TE leg length under 1 sunlight. 31

46 Fig Schematic illustration of manufacturing process for the W-STEG including a process for (a) depositing a Ti/ MgF 2 multilayer solar absorber on flexible PI substrate, (b) dispenser printing of thermoelectric ink on PI substrate with a solar absorber and (c) measuring W-STEG power under AM 1.5G condition. Fig (a) Polyimide substrate with strip shape of Ti/MgF 2-superlattice and (b) printed the TE legs on the substrate Device structure optimization Figure 2.21 shows trends in power characteristics with increasing solar absorber width. The output voltage and power increase from 25.23mV and 1.01μW to 33.26mV and increase to 1.45μW as the solar absorber width increases from 5 to 10mm. However, increasing the absorber width to 15 and 20 mm further increased the output power and voltage to an unalterable level. This means that T of the W- STEG is nearly saturated for widths of absorbent that is greater than 10 mm. The optimal solar absorber width was selected to be 10mm. The output voltage V OC can be calculated as V OC = α ΔT, where α is about 280 μv / K per pair and the absorber width T for 5 and 10 mm are 8.94 and 11.79K.. 32

47 Fig Comparison of voltage and power according to solar absorber width. The arrow in the inserted schematic represent experimental variables. Fig Comparison of voltage and power according to single TE leg length. The arrow represents the experimental variables. As demonstrated in fig. 2.17b, the temperature difference in the transition region near the boundary of the solar absorber is gradual. Therefore, depending on the TE leg length on the side of the PI board at 33

48 the boundary of the solar absorber, ΔT across the TE legs may be different. However, the total internal resistance value of the W-STEG increased as the TE leg length also increased, the optimum length of the TE leg is selected by the output power characteristics and the relative TE leg length. The length of the TE leg was increased from 5 to 15 mm and the output power characteristics of W-STEG with 10 pairs of TE legs was measured (Figure 2.22). The total voltage and resistance increased from 21.36mV and 112 Ω to 34.42mV 141 Ω, respectively, as T increased from 7.57 to K. As leg length increases from 10 to 15 mm, the resistance of the ten pairs of TE legs increased by 50%, however ΔT did not increase as much, reducing the output. As a result, the W-STEG performance was optimized at 10mm Thermal imaging of the W-STEG For visualization of the W-STEG with temperature difference, a self-heating solar absorber (under the condition of AM 1.5 G), infrared image of the working W-STEG was obtained. The highest and the lowest temperature of the TE legs were approximately 46.7 C and 35.3 C, respectively, when the entire device was exposed to radiation, and the output voltage and power were measured at mv and 1.45 μw, respectively. The T at this point was 11.4 K, and this value differs from the temperature gradient formed in the substrate itself with the solar absorber. It was considered that the temperature of the cold side is increased due to the absorption of light by the BiTe legs, heat conduction, and the Joule heating by internal resistance. Heat conduction or Joule heating is inevitable among these various mechanisms, however blocking light can prevent the light absorption of TE legs. Thus, only BiTe legs were blocked from the light to keep the cold side temperature as low as possible. Figure 2.23a and b show the thermal images obtained after shielding the legs with aluminum foil to block the light absorption of the BiTe legs and the corresponding current-voltage-output characteristic curves of the W-STEG. After that, the T dramatically increased to 20.9 K, in which the maximum and minimum temperatures of the TE legs were about 46.6 C and 26.7 C due to the remarkable decrease of the temperature at the end of the TE leg corresponding to the cold side. The output voltage and power were also maximized up to 55.2 mv and 4.44 μw. Also, the estimated T was 11.3 K and 18.4 K, respectively, which are almost identical values obtained from the thermal images. 34

49 Fig Thermographic image of W-STEG (a) with and (b) without light blocking to TE legs taken with infrared camera The characterization of W-STEG Figure 2.24 is a bar graph that demonstrates the output voltage and ΔT of WTEG printed on various substrates such as glass fabric, PI substrate and PDMS using a variety of techniques such as screen printing, pasting and dispenser printing. More information is given in Table 4 [55-58]. All WTEGs excluded from this study are driven by body temperature. If the generator is operated at an ambient temperature of 15 to room temperature, the ΔT of the WTEG is only 1.5 to 4.1 C. The W-TEG fabricated in this study has ΔT of 20.9 C and, as far as we know, the highest ΔT of the WTEG. Fig IV curve of W-STEG and W-STEG with TE legs obscured. 35

50 2.5 Application of WSTEG The process of thermoelectric painting Due to a layered structure of the crystal, BiTe materials are known to be very fragile [59]. In a real world application of wearable TEG, rigidity and flexibility of the device must be compromised. Previous studies have shown that high density semiconducting polymer such as polystyrene sulfonate, such as PEDOT: PSS [60], penetrate the epoxy resin [61] or BiTe surface to insert micron sized BiTe particles. To increase flexibility. However, since epoxy resin is an insulator, it lowers the electrical conductivity, or PEDOT: PSS increases the thermal conductivity and decreases TE performance. In this study, Parylene C was deposited on the surface of the device to a thickness of ~ 1 μm to enhance flexibility of the WTEG without decreasing its thermoelectric properties. The parylene C does not cause thermal deformation of the device since it is conformally deposited on rough surfaces at room temperatures [62]. Parylene C also has a low thermal conductivity of Wm -1 K -1 and therefore an excellent insulator which makes it ideal material for coating of WTEG [63] Flexibility of the W-STEG Figure 2.25 demonstrates the internal resistance change for the bending test to see the effect of parylene C. The bending test was repeated for 150 cycles, and the resistance of a single TE leg was measured at a bending radius of 12 mm to 6 mm. The data showed a notable difference in electrical resistance up to 150 cycles depending on the presence of the parylene C. The resistance of the BiTe legs without parylene C showed a significant increase of electrical resistance by 28% at 12 mm and 141% at 10 mm, and eventually, the BiTe leg failed to withstand the repeated deformation at 8 mm. However, for the sample with parylene C, there is a low variation of 2.9% at 6 mm and 1.5% at 12 mm. From these results, it can be deduced that although parylene C is as thin as 500 nm, it strongly reduces strain and therefore contributing to the minimization of tensile and compressive strain of BiTe leg. Figure 2.26 is a schematic illustration showing a mechanism which shows the improvement of flexibility by parylene coating. As seen in the cross-sectional SEM image of the BiTe leg (Figure 2.27), microcracks are present on the surface of BiTe. With tensile stress applied to the surface, the microcrack ultimately destroys the linkage between the BiTe surface. However, with Parylene C coating of the BiTe surface, the microcracks are filled by Parylene C. The parylene C coating enables the stress to be spread evenly, therefore improving the flexibility of the substrate. 36

51 Fig Internal resistance stability for repeated bending cycle along the different bending radius. Fig The schematic illustration of the stress during bending cycle with / without parylene C. Each stress line shows the stress concentration and stress relaxation. 37

52 Fig The cross-sectional SEM image of the parylene C coated TE legs, which shows the micro crack filling Application of the W-STEG The W-STEG of this study can be easily attached to various exposed surfaces such as clothes for human bodies and windows for buildings. W-STEG was tested in a real environment to see the realworld output voltage produced by the sunlight. The test was executed at ambient temperature of 20 C and a wind speed of 4 m / s. The TE legs were covered to maximize the ΔT between the high and low temperature sides. Under outdoor conditions, the output voltage was measured at 52.3 mv with a ΔT of C, which is very close to the measured output voltage and ΔT using the 1-sun Solar Simulator. Fig Photograph of the W-STEG attached to clothes and windows. Figure 2.29 shows the output voltage monitored over time after the W-STEG is exposed to sunlight. Previous wearable TEGs that are driven by body temperature tend to produce a high output voltage momentarily upon contact with the skin, and the output value decreases over time due to thermal equilibrium. 38

53 Fig Comparison of wearable TEGs application results, voltage and T. However, this W-STEG maintains the initial output voltage after reaching a steady state within ~ 20 seconds. These results are expected to be used as self-running technology of wearable electronic devices by promoting the application of WTEG in the early stage of development. 39

54 Fig Voltage-time curve showing immediate response to light and retention of the voltage over time The optimum temperature difference considering the heat flux To optimize ΔT, we covered the hot side with PDMS to block convection of air to hot side. For our heat sink material, we used porous-copper foam to actively generate heat flow and promoting convection (Fig. 2.31). It can be expected to increase of the device efficiency by securing the maximized structure of temperature difference considering heat flux. Fig Schematic illuttration of the strategies to enhance temperature difference. We need to consider the heat flux and thermal resistance to understand the temperature gradient in the structure. When the heat flows and ΔT is formed through a plane wall, heat transfer occurs by convection from the hot side to cold side through the planar substrate. Heat transfer is indirectly proportional to the length of the thermal circuit. And also, thermal resistance can be expressed as R t = l. Lower thermal conductivity would result in the increase of resistance which essentially increases the ΔT. The following, ka 40

55 in terms of structural conditions corresponding to the cross-sectional area and length in the eq. of thermal resistance. Depending on the direction of thermal flow, the cross-sectional area and length can be expressed as follows. When the heat flows in the horizontal direction, the cross-sectional area is relatively small and the length is relatively large as compared with the vertical direction, it caused larger thermal R, and larger ΔT. Therefore, higher heat flux per unit area and thermal resistance contributes to the increase of ΔT which implies that it is more efficient to form a temperature gradient on a planar structure. Based on this principle, the air trapped in the PDMS cover has the effect of increasing the hot-side temperature up to 5 C as a medium with low thermal conductivity by application of insulation with PDMS cover. In addition, by applying porous copper foam to more actively generate convection, the temperature of the cold side, which is edge side of the device, could be lowered by 5 C. Fig Comparison graph of the temperature difference according to wtepwise strategy for increasing temperature difference in a planar substrate. 2.6 Conclusion In summary, I proposed a new way to surmount the low ΔT problem encountered by general WTEGs. This is one of the major disadvantages of WTEG technology in which heat-induced from surface of the human body. The proposed wearable solar TEG can generate a high value of ΔT to ~ 21 C by a locally deposited solar absorber on the substrate. The solar absorber is a five-period consisting of Ti / MgF 2 multilayer structure, where each layer designed by optimum dimension so that the structure effectively reduce the reflectance and induce the absorption by interference effects. The thicknesses of each layers 41