A Quasisolid Electrolyte Thermogalvanic Cell by Using Sand Grains

Thermogalvanic cells (TGCs) can convert low‐grade heat into electric energy. Currently, Carnot efficiency of the P‐type TGC can reach higher than 11%, but the TGC's commercialization is still limited due to the low Carnot efficiency of the N‐type TGC. In this paper, new TGCs are reported, and they are made of desert sand grains mixed with the Cu(NO₃)₂ aqueous electrolyte, named as QS‐Cu‐TGCs, the thermoelectric performance of which is investigated. It is found that although sand grains are common, widely seen in the desert, the Carnot efficiency of such N‐type QS‐Cu‐TGCs is quite high, about 4%, with the Seebeck coefficient of 8.4 mV K−1, driven by the temperature difference 12.1 K between two electrodes, in an environment below 373 K.


DOI: 10.1002/aelm.202300089
N-type (P-type) cells. N-type and P-type cells form P-N junctions, which are connected in series into commercial application modules. [12] Currently, there are remarkable results of P-type TGC. [10,[13][14][15] It is found that the cell's Seebeck coefficient can be enhanced by thermosensitive crystallization-induced redox ion concentration gradient, which causes the Carnot efficiency ( r ) of P-type TGCs enhanced to 11.1%. [16] However, the research on Ntype TGCs seems a bit slow. [17][18][19][20][21][22][23] Kim et al. propose an N-type TGC based on Fe 2+ /Fe 3+ redox couple with a r of 0.53%. [24] Yu et al. develop an N-type TGC based on Cu/Cu 2+ redox couple with a S e of 1.66 mV K −1 and a power density of 3.5 W m −2 by using 3D multistructured electrodes, which make it possible to work together with the highperformance P-type cell in a P-N junction, [25] and the r of this cell is expected to be reported. That means the commercial module of TGCs is mainly limited by the thermoelectric performance N-type TGCs. [26,27] Have you ever been enchanted by the neat sand ripples or shocked by the tall sand mountains in the desert? Do you know? The basic units that make up these landscapes, the sand grains, actually carry net electrical charges. [28,29] The sign of charges carried by sand grains is related to the PH value of surrounding electrolyte solution. By ion adsorption method, we measure the net surface charges of sand grains at the southern edge of the Tengger Desert, China. With the electrolyte solution PH < 3.3 (PH > 3.3), the surface of the sand grain has a net positive charge (negative charge) [30] to see Figure S1 (Supporting Information). The sand grains mixed with the Cu(NO 3 ) 2 aqueous electrolyte are inserted between two parallel copper electrodes to make a sandwich-structure TGC, called the quasisolid Cu/Cu 2+ redox couple thermogalvanic cells (QS-Cu-TGCs). Both of N-type and P-type cells can be made, only by changing the concentration of the Cu(NO 3 ) 2 aqueous electrolyte. Charged sand surface leads to a strong Soret effect, which further enhances Seebeck coefficient. The largest r , 4%, of N-type QS-Cu-TGCs can be reached at ΔT = 12.1 K, while the S e of P-type QS-Cu-TGCs reaches −17.3 mV K −1 . The open-circuit voltage of 10 P-N junctions connected in series can reach to 2 V at ΔT = 11 K, which can power the electrical watch, etc. , and c 1 ) are the schematics of the Cu-TGC, N-type Qs-Cu-TGC with PH<3.3 and P-type Qs-Cu-TGC with PH>3.3. a 2 ), b 2 ), and c 2 ) are the schematics of ions motion in the bulk solution or diffuse layers of cells of a 1 ), b 1 ), and c 1 ). a 3 ) The thermal voltage (ΔV e ) of Cu-TGC is mainly composed of the redox couple thermal voltage (ΔE 0 ), and the thermal diffusion voltage (ΔV td ) can be neglected, that is ΔV e ≈ΔE 0 . b 3 ) The ΔV e is derived from the synergistic effect of ΔE 0 and ΔV td . c 3 ) The ΔV e is derived from the antagonistic effect of ΔE 0 and ΔV td . Cu(NO 3 ) 2 aqueous electrolyte with different concentration. The sand grains with a diameter of 0.22-0.3 mm are selected to be mixed with the Cu(NO 3 ) 2 aqueous electrolyte to be the quasisolid electrolyte. The electrode is made of copper foil glued to a methyl methacrylate (PMMA) plate, and the thermocouple probe is pasted between the copper foil and the PMMA plate to measure the temperature of the electrode. A short PVC tube connects with electrodes on the copper foil side, and then the solid electrolyte is pressed into the tube fully, sealed with other electrode, the copper foil side contacting with the electrolyte. Finally, by using hot-melt adhesive the sandwich structure is sealed, and a QS-Cu-TGC is made. When the copper foil is replaced by graphite paper, QS-TC is made. Driven by temperature difference (∆T) between the two copper electrodes, a thermal voltage difference (ΔV e ) between two electrodes immersed in the Cu(NO 3 ) 2 aqueous electrolytes can be produced, which is called an aqueous electrolyte Cu/Cu 2+ redox thermogalvanic cell (Cu-TGC), [9,[31][32][33][34][35] proposed by Bouty in 1880. [35] The Cu-TGC is a typical N-type TGC. Oxidation (Cu-2e→Cu 2+ ) occurs at the cold electrode and reduction (Cu 2+ +2e→Cu) occurs at the hot electrode (Figure 1a 1 ). ΔV e of Cu-TGC mainly consists of the redox thermal voltage ΔE 0 = R ΔT and the thermal diffusion voltage (ΔV td ), ΔV td = S td ΔT. [15] The R is the temperature coefficient of the standard electrode potential, and S td is called the thermal diffusion potential, which is due to the Soret effect (Figure 1a 2 ). [11,15] The R of Cu-TGC is ≈0.5-1.3 mV K −1 , [31][32][33][34] and S td is usually negligible (≈10 μV K −1 ) in aqueous solutions. [15] Therefore, ΔE 0 is dominant in ΔV e of Cu-TGC, that is ΔV e ≈ΔE 0 (Figure 1a 3 ). By adding sand grains into the Cu(NO 3 ) 2 aqueous electrolyte of Cu-TGC, a new cell can be made, and it is called a quasisolid electrolyte Cu-TGC (QS-Cu-TGC). The PH value of Cu(NO 3 ) 2 aqueous electrolytes with different concentration is shown in Figure S1 (Supporting Information). When QS-Cu-TGCs are made with high concentration Cu(NO 3 ) 2 aqueous electrolyte (PH < 3.3) to see Figure 1b 1 ), and the surface of the sand grains is positively charged, that means more Cu 2+ ions are adsorbed in the stern layer, so that the concentration difference between NO 3 − ions and Cu 2+ ions become large in the diffuse layer. The zeta potential of sand grains increases as the temperature increases, [36] and thus at hot end, more Cu 2+ ions are absorbed in the stern layer than the ones at cold end. Therefore, the concentration difference between the Cu 2+ ions and the NO 3 − ions in the diffuse layer at hot end is larger than the that at cold end. NO3 − ions in the diffuse layer at hot end is much higher than that at cold end, which leads to the more NO 3 − ions move toward to the cold end, which forms ΔV td (Figure 1b 2 ). Both ΔV td and ΔE 0 act synergistically to enhance ΔV e (Figure 1b 3 ). When the QS-Cu-TGC is made with low concentration Cu(NO 3 ) 2 aqueous electrolyte (PH > 3.3) (Figure 1c 1 ), the surface of the sand grains is negatively charged, that means NO 3 − ions are absorbed in stern layer, and it also causes the concentration difference between NO 3 − ions and Cu 2+ to become large in the diffuse layer, however Cu 2+ ions at the hot end are much more that the ones at the cold end, so that driven by the temperature, the more Cu 2+ ions at hot end will move toward to the cold end. The ΔV e of QS-Cu-TGC is derived from the antagonistic effect of ΔE 0 and ΔV td (Figure 1c 2 ). When |ΔV td |>|ΔE 0 |, the electric potential at the cold end is higher than that of the hot end, so that the QS-Cu-TG is a P-type cell (Figure 1c 3 ). The principle of QS-Cu-TGCs is different from the one reported, [37] in which nanoparticles added into electrolyte solution can move to the cold end driven by the temperature difference. Because the nanoparticles adsorb the electrolyte ions, ionic thermodiffusion is enhanced. In this paper, the sand grains cannot move driven by the temperature difference, however the concentration difference of two ions (Cu 2+ and NO 3 − ions) in the diffuse layer can be formed due to the sand grains surface selectively adsorbing ions, and it further enhances the concentration difference of same ions between hot end and the cold one because the zeta potential increases as the temperature increases for sand grains, and thus the more Cu 2+ /NO 3 − ions move toward to the cold end to produce the potential difference as mentioned above. However the enhancement in the voltage of both the cells [37] and QS-Cu-TGCs is due to the increase of the ions concentration difference driven by the temperature difference.

Experimental Results and Discussions
The thermoelectric performance of typical QS-Cu-TGCs is shown as in Figure 2, in which Cu(NO 3 ) 2 concentration is 2.5 m, and PH = 1.7, and the mass moisture content is 9%, unsaturated shown as in Figure S3c (Supporting Information). The crosssectional area of the cell is 2.54 cm 2 , and the distance between the two electrodes is 1.5 cm. The V oc increases as ΔT increasing shown as in Figure 2a by changing the hot electrode temperature (T h ). The cold electrode temperature (T c ) is kept at ≈315 K, and ΔT = T h -T c ). The results of QS-TCs and Cu-TGCs are also shown in Figure 2a. It is shown that V oc of both QS-Cu-TGCs and QS-TCs increases rapidly as ΔT increases, while V oc of Cu-TGCs increases slowly as ΔT increases. Here QS-TCs are same as QS-Cu-TGCs, just with the copper electrodes of QS-Cu-TGCs replaced by graphite paper. Thus driven by temperature difference, there is no redox reaction at two ends in QS-TCs, which is also validated by rectangular CV curve to see Figure S5 (Supporting Information). That means V oc of QS-TCs does not include ΔE 0 , so that ΔV e = V td . From Figure 2a, it can be found that the V oc of QS-Cu-TCs, QS-TCs, and Cu-TCs all increases linearly with ΔT increasing, and the slope of the V oc −ΔT curve is calculated cells' thermal potential (S e ), which is 10 mV K −1 for QS-Cu-TGCs, while it is 8.3 mV K −1 for QS-TCs and 0.89 mV K −1 for Figure 3. a) r and ZT reported in references and in this paper, and b) performances of QS-Cu-TGCs and cells reported, [9] detailed in Table S1 (Supporting Information).
Cu-TGCs, calculated by formula (S1). Because of lacking thermocouple effect, the thermal potential is only the thermal diffusion potential (S td ) for QS-TCs, while the thermal potential is only the thermocouple potential ( R ) for Cu-TGCs, without thermal diffusion effect. Therefore, we can calculate the rate of S td and R in S e of QS-Cu-TGCs, which are 83% and 8.9%, respectively, and the rest is 8.1% caused by other factors to see Figure 2b. The V oc of QS-Cu-TGCs is significantly higher than that of Cu-TGCs at the same temperature difference. It indicates that the charged sand surface causes a strong Soret effect to produce the dominant potential in ΔV e of the QS-Cu-TGC. We also investigate the electrical conductivity ( ), thermal conductivity ( ), and figure of merit (ZT) of QS-Cu-TGC and Cu-TGC at different average temperatures (T ave ) (T ave = (T h +T c )/2) shown as in Figure 2c-d, and the detailed measurements and formulas to see formulas (S2)-(S4) in the Supporting Information. From Figure 2c, it can be inferred that the sand grains impede thermal convection to reduce the thermal conductivity, and also reduce the electrical conductivity, however, the reduction of is much higher than that of , which results in that ZT and r of QS-Cu-TGCs are significantly higher than that of Cu-TGCs at ΔT = 12.1 K (T ave = 323 K). The ZT of QS-Cu-TGCs is about 0.2 (Figure 2d), and the r reaches 4%, which is nearly 50 times higher than that of Cu-TGCs (Figure 2e). r of QS-Cu-TGCs is the highest compared with ones of N-type cells published in other work to see Figure 3; and Table S1 (Supporting Information) in detail. The Normalized maximum power density (P max /(ΔT) −2 ) of QS-Cu-TGCs reaches a maximum value of 0.72 mW m −2 K −2 at ΔT = 9 K (Figure 2e). The thermoelectric performances of QS-Cu-TGCs with different moisture content are shown in Figure S7 (Supporting Information). V oc and short circuit current (I sc ) of the QS-Cu-TGCs are shown in Figure 2f for 40 h operation at the ΔT of ≈9 K (T c = 315 K), that means the QS-Cu-TGCs can output stable power for long time.
From Figure 3, it can be found although QS-Cu-TGCs driven by lower temperature difference than the one reported, [9] QS-Cu-TGCs perform well in the thermoelectric side. In addition, the electrodes used [9] is Pt, which is expensive.
36 N-type cells (cross-sectional area of 7 cm 2 , the distance of 1.5 cm between the two electrodes) are connected in series as a module to demonstrate viability for scale-up shown as in Figure 4a. The V oc of this module can reach to 2 V at ∆T ≈ 13 K (Figure 34b). According to the voltage-current curve and voltagepower curve, the maximum power of this module reaches 240 μW to see Figure 4c). The module can directly drive various electronic devices, including LED lamp beads and electronic thermometers (Figure 4d,e).
Effect of the Cu(NO 3 ) 2 concentration in QS-Cu-TGC on V oc −ΔT is shown as in Figure 5a), from which it can be found   Figure 5b at ∆T = 10 K. As shown in Figure 5b, S e reaches 14.3 mV K −1 when the concentration is 1.5 m, while S e reaches −17.3 mV K −1 when the concentration is 0.1 m. However, although S e is large with the low concentration of 0.1 m, the low concentration results in the low conductivity, which causes the low power density of the P-type TGC to see Figure 5c. According to the voltage-current curve and voltage-power curve of the Ntype QS-Cu-TGC and P-type QS-Cu-TGC, the maximum power density of the N-type cell is 56 mW m −2 , while that of the P-type cell is only 1 mW m −2 (Figure 5c).
The V oc of P-N junction in series is enhanced, higher than that of single P-type cell or single N-type cell ( Figure S8, Supporting Information). 10 groups of P-N junctions are connected in series ( Figure S8a, Supporting Information), and the V oc can reach to 2 V at ΔT = 11.7 K ( Figure S8b, Supporting Information). However, due to the limitation conductivity of the P-type cells, the maximum power of the 10 P-N junctions is only 19.3 mW ( Figure S8c, Supporting Information), which can only power an electronic watch ( Figure S8d, Supporting Information). Fortunately, there are promising P-type TGCs with high ZT value as mentioned in Introduction. One can connect current P-type cells with high ZT such as the ones reported [16] with QS-Cu-TGCs in series to make P-N junction to pursue high power. In addition, it was found that P-type TGCs can hold the voltage for long time such as 100 h with high Seebeck coefficient, −17.3 mV K −1 , to see Figure S9 (Supporting Information).

Conclusions
In this paper, both of P-type and N-type QS-Cu-TGCs are made by changing the concentration of Cu(NO 3 ) 2 aqueous electrolyte mixed with sand grains. The Seebeck coefficient (8.4 mV K −1 ) and Carnot thermal efficiency (4%) of the N-type TGC are promising. P-type TGC has a high Seebeck coefficient of −17.3 mV K −1 , and a very strong voltage maintenance capability. In addition, plat-inum electrodes used in most TGCs, compared with which copper electrodes used in QS-Cu-TGC are element-rich and easy-tomake (the price is only 1/5300 of platinum electrodes [25] ), which requires relatively low cost. The production process is more conducive to commercial application. In addition, the sand grains are environmentally friendly.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.