Power Generation at Room Temperature -How to Design of the Sensitized Thermal Cell-Haruki

The effective utilisation of thermal energy is crucial to a world aiming at sustainable development goals. The sensitised thermal cell (STC) is a new thermal energy conversion technology, which was reported in 2017, for generating electricity via the redox reactions of electrolyte ions with thermally excited carriers in semiconductors. STC is attracting attention as a technology that could affect oil prices globally. Here, we report the successful construction of STC, which discharges and recovers well at room temperature, by studying the interelectrode and ion diffusion distances. The fabricated STC is thinner than 0.5 mm and can be rendered exible. Further, STCs, which possess a power generation capacity of 0.1 mA, 100 mV per 1 mm 2 area, and shall be installed “in” internet-of-things (IoT) devices, drainage pipes, and walls in the future.


Main Text
It is globally challenging to develop technologies that can control CO 2 emissions without reducing the energy supply. The heat energy that is all around us is an attractive green energy source. [1][2][3] Geothermal power generation 4 and the Seebeck effect 5 are the two main methods of generating electricity from heat. Regarding the former, electricity is generated via the rotation of a steam turbine, in which the heat used to generate the steam is supplied by a geothermal source. A major setback of this method is the availability of suitable land since the water demand is high. In the latter, electricity is generated via a temperature difference, which induces an electric potential. The device consists of at least two dissimilar conductors, and a complex design is required to ensure that it works under all possible conditions. To solve these problems, a new power generation method that could directly convert heat into electricity, therefore eliminating concerns with radioactive waste and enabling the use of renewable energy to generate electricity with a low environmental load using low-cost heat, is desirable.
We recently reported a sensitized thermal cell (STC) that is a new thermal energy conversion system [6][7][8][9][10][11] , which was inspired by the concept of a dye-sensitized thermal cell (DSSC). [12][13][14][15][16] Utilising STCs, electric power can be generated "directly" from heat via the redox reactions of electrolyte ions with thermally excited carriers in semiconductors (Fig. 1). 6,7 Since this battery can function at a constant temperature, we rst expected that the generation of electricity would stop at some point when equilibrium is reached.
attained. However, we surprisingly observed that the reaction could be restarted by the simple ip of an on/off switch in the external circuit. The discovery of the restart phenomenon means that STC is a technology that directly converts heat into electricity, which could affect the price of oil.
In this work, we examined the cause of this restart and acquired a guideline for designing batteries that can generate power at the desired temperature. As a result, we successfully reduced the power generation temperature from 80 °C 17 to room temperature (RT, 30 °C). The fabricated STCs were thinner than 0.5 mm and can be rendered exible.
What we rstly examined STC possesses a simple layered structure with a working electrode containing a semiconductor and a counter electrode sandwiching an electrolyte (Fig. 2a). To examine the "Restart" phenomenon, we focused on the distance between the working and counter electrodes. We fabricated 4 different cells with interelectrode distances of 85, 114, 228, and 342 μm, respectively. The obtained cells were named according to their corresponding electrode distances (Cell-85, -114, -228, and -342). The employed electrolyte volume was proportionally based on the interelectrode distance to eliminate the difference in the electrode/electrolyte contact area between each cell. A Ge-based semiconductor and a copper ionbased polyelectrolyte were selected for the STC materials. [9][10][11]17 The fabricated STCs demonstrated stable power generation at 80 °C (Supporting Information (SI) 1). The open circuit voltage (V OC ) of Cell-85, -114, -228, and -342 were approximately the same at 362, 373, 399 and 406 mV, respectively. The V OC value is caused by the difference between the redox potential of the electrolyte ions and the Fermi level of the working electrode. 10 Comparing the discharge capacities of the 1st 200 nA discharge, that of Cell-342 was the smallest, thereby indicating that the sustainability of the power generation was in uenced by ion convection inside the electrolyte.
After the initial discharge, all the batteries demonstrated voltage recoveries when they were switched off (SI 2). However, the time required to stabilise the voltage increased with increasing interelectrode distance. Cell-85 recovered up to 0.35 V in few tens of seconds, while Cell-228 and -342 recovered up to 0.25 V in 4 h, and this indicates that the interelectrode distance could affect the restart of STC. Notably, all the batteries were left in the thermostatic bath during discharge and restart processes.
The 200 nA discharge characteristics of each battery at 80°C were tested 4 times. In Cell-228 and -342, the V OC after the 1 st discharge was 100 mV smaller than the initial V OC (Fig. 2b). Contrarily, in Cell-85 and -114, the V OC attained the initial value after each discharge.
The 2 nd and 3 rd discharge curves are shown in Figs. 2c and 2d, respectively. The voltage rapidly dropped to zero in Cell-228 and -342. Meanwhile, the 2 nd discharge capacities in Cell-85 and -114 were 0.21 and 0.24 mAh/g, while the 3 rd discharge capacities were 0.34 and 0.08 mAh/g, respectively. It is noteworthy that the discharge capacity of Cell-85 was higher than that of Cell-114.
These results strongly indicated that ion convection was crucial to the restart phenomenon (the cell can restart if su cient ions are supplied to the electrode interface). Here we should mention that the 2 nd and 3 rd discharge capacities dropped more than the 1 st discharge. We think this is because of the polyelectrolyte 18,19 . During the rst discharge, the electrolyte was not completely mixed at the molecular level, thus causing the diffusion of ions via local concentration distribution. Conversely, in the 2nd and 3rd discharge processes, power was generated after the complete mixing of the electrolyte in the rst discharge, resulting in the shorten discharge time. These results suggested that ion diffusion in the electrolyte is re ected by the restoration of power generation by STC.

Production of thinner cells
Since the maximum characteristics were obtained with the thinnest cell (Cell-85) among the tested ones, we applied comb-shaped electrodes to assess what would happen if the interelectrode distance was narrower. Thus, we fabricated Ge and Pt comb-shaped electrodes on quartz, glass and a Kapton tape (Fig.   3a).
Two interelectrode distances (2 and 5 mm) were examined. The cells were referred to as Cell-2 and -5. These comb-shaped electrodes were lled with the electrolyte, and the battery characteristics were measured on a hot plate while controlling the temperature.
Surprisingly, Cell-2 did not generate power at 80 °C (Figs. 3b and c). However, it generated power well as the temperature was lowered, and the best battery characteristics were obtained at 30 °C, which was near RT (SI 3a). We were so surprised that we also tried to generate power at room temperature (RT, ~25 °C under air conditioner), and con rmed that the power generation capacity (270 mV, Fig. 3d) was stable at least over 7000 s. (Fig. 3e). This power generation and restore experiments will continue for several days, and the room temperature will change during that long experiments. In order to eliminate the effect of this temperature change, the following results were measured at 30°C xed by a hot plate.
Cell-5 exhibited the same trend; it achieved good voltage recovery even after repeated discharge cycles (Fig. 3f). Particularly, the discharge time of Cell-5 at 30 °C was longer than that of Cell-2 (Fig. 3g). To understand the phenomenon, we measured the cyclic voltammetries (CVs) of Cell-2 at different scan rates at 30 °C and 50 °C. We observed a capacitor-like behaviour at 50 °C (SI 3b and 3c). These results indicated that extremely short or long interelectrode distances are not desirable for restarting STC and that there is a 'just right' interelectrode distance.

What happened?
These results suggested that the ion diffusion length from the electrode surface affected the generation of power. Thus, we measured the ion diffusion length using the AC impedance measurement 20 (SI 4). The results of the temperature change revealed that the ion diffusion length increased at higher temperatures (Fig. 4a). Conversely, the ion diffusion lengths of Cell-85, 114, 228 and 342 increased with decreasing interelectrode distances at 80 °C (Fig. 4b). During the power generation, a potential difference, V, was generated between the working and counter electrodes. The magnitude of the electric eld, which was applied to the entire electrolyte, was equal to V divided by the interelectrode distance. Resultantly, the electric eld was larger with a shorter interelectrode distance, which made it easier for electrolyte ions to diffuse, thereby increasing the ion diffusion length.
From these measurements, it was con rmed that power generation occurred when the value S (the interelectrode distance divided by the ion diffusion length) was between 1 and 10. The same measurements were conducted for other STC systems, and power was generated at 1 £ S < 10 (SI 5).
Based on these results, we considered the power generation and restore mechanism of STC ( Figure 5). At S < 1, the electrolyte ions behaved as a capacitor and did not move from the electrode interface. Hence, power was not generated (Fig. 5a). When S is too large, the electric eld became small and the ions, which were generated at the Ge electrode (or counter electrode) interface did not diffuse to the counter electrode (or Ge electrode) interface and there was no restart (Fig. 5c). Put differently, STC generated and restored power well at 1 < S < 10, where ion diffusion constantly occurred.

Now we can design favourable STCs
The ion diffusion length is a value that is determined by the electrolyte material and temperature. That is, if you measure the ionic diffusion length of the electrolyte at the temperature where you want to install the STC, and design the interelectrode distance to be several times longer than that, you can obtain an STC that would generate and restart power well. Since STC can be 'buried' in heat, it has been applied to operate liquid crystal displays at 80 °C 9 , Bluetooth communication devices on asphalt in the summer (SI 6). In principle, V OC of STC is independent of the temperature and is the difference between the Fermi level of the working electrode and the redox level of the electrolyte ions. 10 Regarding Ge/(Cu ions in polyethylene glycol (PEG)) STC in this work, V OC was ~0.35 V. Assuming that the theoretical short-circuit current value was the number of reactive ions reaching the electrode interface per second, it would be estimated as ~60 and 150 mA/cm 2 at RT and 80 °C, respectively, for the Ge/(Cu ion in PEG) STC combshaped electrode (5 mm) based on the saturation concentration of the electrolyte, the ion mobility 21 and electric eld that were applied to the electrolyte (SI 7). This indicated that the output powers were 0.2 and 40 mW/cm 2 at RT and 60 °C, respectively, which are comparable to that of solar cells (6 mW/cm 2 ) and are su cient to employ STCs in IoT devices (sensor: mW~, wireless power transmission: mW~) 22 .
Evidently, these theoretical values were 'instantaneous' maxima, and the discharge and restore times would be the other major parameters to be considered for the practical application of STCs.
We strongly expect our ndings to offer huge energy markets for industries and academia as well as help achieve the control of CO 2 emissions without reducing the energy supply, thereby contributing towards solving global environmental issues.

Materials And Methods
Wafer electrodes. The electrodes were composed of an n-Si/Ge substrate (Ge layer: 2 μm) that was manufactured by Tohnic Corporation, Japan. A Cr layer (20 nm) was sandwiched in the n-Si wafer (0.02 Ω cm, 15 × 20 × 0.5 mm) to improve the adhesion of the n-Si/Ge interface. The resistivities of the n-Si and Ge sides, which were measured by the 4-probe method (Loresta GP MCP-T600, Mitsubishi Chemical Corporation) at RT, were 3.18 and 3.27 Ω/sq, respectively. The counter electrode was composed of uorine-doped tin oxide (FTO, 10 Ω, 15 × 25 mm).
Comb electrodes. A comb electrode with 2 materials, Pt and Ge, was used (Tohnik Co., Ltd.). Ge (1 μm thickness) and Pt (30 μm thickness) were deposited on a synthetic quartz substrate with a thickness of 0.525 mm and an area of 1 cm 2 . The comb electrode widths were 2 and 5 mm for the electrode distances of 2 and 5 mm, respectively, and the area of the comb-shaped electrode part was 0.04 cm 2 . For the demonstration, we also fabricated the comb electrodes on a kapton tape.
Electrolytes. Since Ge is a water-labile material, the electrolyte was prepared in a ow-type glove box, coupled with a circulation/puri cation machine, lled with Ar (the oxygen concentration was 0.2̱ -0.5 ppm). Before the reagents were used in the glove box, they were vacuumed at 80 o C for 15 h in a side box. Furthermore, the molecular sieves (Fuji lm Wako Pure Chemical Industries, Ltd.) were heated for 3 and 4 h at 180 and 350 °C in an electric furnace, respectively, after which they were cooled to ~50 °C in a vacuum desiccator before they were added to PEG600 to approximately one-third of its original volume for dehydration.
After introducing the reagent into the glove box, the concentrations of CuCl, CuCl 2 and LiCl were weighed 0.5, 0.5 and 0.6 mmol/g (PEG), respectively. The components were mixed for 10 min utilising a mortar and a pestle.
The thermal analysis of the electrolyte was performed by thermogravimetry-differential thermal analysis (TG-DTA). In this analysis, 7.28 g of the electrolyte was heated from RT to 180 °C in an N 2 ow and at a heating rate of 1 °C/min. The thermal behaviour of the electrolyte was compared with that of PEG (the reference material), and three samples comprising PEG and CuCl, PEG and LiCl and PEG and CuCl 2 were selected to investigate the in uence of each reagent.
Cell assembly. The FTO substrate was ultrasonically cleaned with ethanol for 7 min, and a double-sided insulating tape with a hole (6 mm dia.) was attached to it. 4 electrode distances (85, 114, 228 and 342μm) were achieved using two types of insulating tapes (a double-coated Kapton Polyimide Film P-223, Nitto, Japan, thickness: 114 μm and an adhesive tape API214A-25X10, Chukoh, Japan; thickness: 85 μm). The n-Si/Ge substrate was immersed in a solution of 5% HF for 5 min, after which it was rinsed three times with ion-exchanged water for 1 min to remove the oxides on the surface. The FTO and Ge substrates were inserted in the glove box in which all the cell-fabrication processes were performed.
Thereafter, the copper-ion-based electrolyte was proportionally dropped into the hole of the insulating tape according to the thicknesses of the tapes to eliminate the difference in the contact area of the electrode/electrolyte between each cell. Therefore, 0.78, 1, 2 and 3 μL of the electrolyte were dropped when the thicknesses of the tape were 85, 114, 228 and 342 μm, respectively. Afterwards, the n-Si/Ge substrate was placed to cover the hole. The obtained cells were named according to their corresponding electrode distances (Cell-85, -114, -228 and -342).
Performance of the sheet-type cells. The following measurements were performed on each cell at an environmental temperature of 80 °C: CV was measured with a VSP-300 instrument (Bio-Logic) in a system composed of two electrodes. The scan rate was 10 mV/s, and the sweep direction was oppositive than that of V OC . Figure 1 Mechanism of STC. The thermally excited electric carriers reduce and oxidise the electrolyte ions to generate electric current at a constant temperature

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Bluetoothrev.mp4