Boosted thermogalvanic thermopower upon solid-to-liquid phase transition

Abstract

Thermogalvanic cells offer scalable low-grade waste heat recovery using tunable electrode-dependent thermopower and electrolyte-dependent thermal conductivities. However, the use of single-phase electrodes thermodynamically curbs the entropy difference, limiting the thermopower enhancement. Here, we show that phase transforming electrodes achieve significantly enhanced thermopower using the melting phase transition of bulk Na_xK alloys. Under both temporal and spatial temperature gradients, the electrodes exhibit significantly increased thermopower up to 26.1 mV K⁻¹ across the melting point and the generated voltages of 261 mV under 10 K temperature gradient. We also show that stabilizing the liquid metal electrode–electrolyte interface plays a critical role in evaluating the thermopower associated with the phase transition. The strategies demonstrated in this work suggest potential design guidelines towards optimizing thermogalvanic cells to specific temperature ranges.

---

Broader context

Despite the ubiquitous presence of waste heat, an efficient means of low- to medium-grade heat energy harvesting remains to be developed. Thermoelectric generators remain limited in energy conversion efficiency (zT < 1.8) for operating temperatures below 200 °C. Also, large thermal conductivity quickly results in thermal equilibration, reducing the efficiency even further. While thermogalvanic energy harvesters (TGHs) received significant attention as a viable option for alternative thermal energy harvesting, the materials innovation for thermopower enhancement has been largely limited to single phases (either liquid or semi-solid), excluding the potential of dynamic phase transition-induced thermopower enhancement. In this work, we propose thermogalvanic energy harvesters that operate across melting points of the electrode materials, utilizing solid-to-liquid phase transition and thus achieving significantly enhanced thermopower values.

Introduction

Current energy technology achieves approximately 35% efficiency in terms of stored energy-to-work conversion.¹ The remaining 65% of energy is rejected as heat at various temperatures. A majority of this rejected heat occurs below 200 °C, where constructing a thermodynamic cycle for heat recovery remains inefficient.² Thermogalvanic energy harvesters (TGHs) are an alternative candidate for harvesting wasted heat by exploiting the temperature-dependent electrode potentials.^3,4 Upon exposure to a temperature gradient, the two electrodes composing a TGH generate an electrochemical reaction. The electrochemical reaction proceeds such that the hot side electrode decreases in entropy while the cold side increases in entropy. This entropy contribution to the free energy generates the temperature gradient-induced voltage output. The fact that electrochemical reactions occur at the electrodes distinguishes the TGHs from thermophoresis⁵ (also known as the Soret effect⁶) or thermally chargeable capacitors where capacitors often employing liquid electrolytes or ionic gelatins get charged by temperature gradients.⁷ The use of solid-state electrodes in TGHs leads to a potentially larger energy density over thermophoresis or surface-dominated thermal capacitors.

TGHs possess unique advantages over the established thermoelectric analogs. Despite the recent advances including metallic thermoelectrics⁸ or flexible thermoelectrics,^9,10 improving the thermoelectric efficiency often expressed in terms of the dimensionless figure of merit (zT) involves constrained optimization of the mutually coupled physical properties.¹¹ The thermopower for thermoelectric materials (Seebeck coefficients, S) is inversely correlated with electrical conductivity, making it difficult to achieve a high power factor (σS²). The thermopower for TGHs (α), however, is dominated by the isobaric partial molar entropy at the electrodes and this relationship has also been exploited to probe the structural characteristics of solid electrodes.¹² Since the transport properties (both ionic and thermal) are often limited by liquid electrolytes, the decoupled behavior among thermopower and transport properties creates a unique opportunity in enhancing the power factor of TGHs. Furthermore, liquid electrolytes employed in TGHs result in increased thermal resistance, impeding thermal equilibration within the generator.

Despite research into thermogalvanic devices, the overall efficiency of thermogalvanic cells remains notably limited.¹³ The overall electric conductivity of the device, dominated by the ionic conductivity, is inherently lower than the electronic conductivity of thermoelectric solids.¹⁴ Since the power factor scales to the square of the thermopower, increasing thermopower among TGHs may overwhelm the limited ionic conduction. Various redox-active materials have been studied in terms of thermopower among both solids and dissolved ionic electrodes. Linford et al. employed various optimized Li-battery electrode materials in TGHs and the observed thermopower lies within the 1–3 mV K⁻¹ window for most of the solids.¹⁵ Dissolved ionic species such as Fe(CN)₆^3−/4− have also been researched extensively, exhibiting a boosted thermopower of 4.2 mV K⁻¹ with additives that induce thermosensitive crystallization¹⁶ or solvent reorganization around the redox molecules.¹⁷ Temperature-dependent ionic gelation among polarized electrolytes has also been studied to yield a high thermopower of 9.6 mV K⁻¹.¹⁸ Other devices have aimed to exploit two distinct physical mechanisms such as thermogalvanic and thermodiffusion in ionic gelatins¹⁹ or thermodiffusion and thermally charged capacitance,²⁰ demonstrating markedly high thermopower values. It is noted that thermally charged capacitors, as categorized by a recent review,¹³ report notably high thermopower above 30 mV K⁻¹. Among thermogalvanic devices, however, most advances have been limited in terms of materials selection, optimizing either dissolved ionic species (e.g. ferricyanide or iodide ions) or ionic gelatins. While dense electrodes potentially provide a high energy density and decoupled figure of merit, the thermopower enhancement among the electrodes has been stagnant and limited in terms of the design principles. The similar thermopower values of a diverse set of solid- and dissolved-state electrode materials imply that the entropy generated upon an electrochemical reaction lies on the same order of magnitude, requiring a novel mechanism for thermopower enhancement.

Here, we demonstrate significantly enhanced thermopower with bulk metal alloys undergoing melting/solidification phase transitions. While traditional thermogalvanic cells rely on redox reactions facilitated by redox couples involving dissolved ionic species, this study shows that high thermopower can be achieved through thermogalvanic cells that leverage changes in the chemical potential of materials themselves with temperature and entropy changes during phase transitions, without relying on the oxidation state changes of transition metal ions. Based on the thermodynamic rationale that thermopower is directly proportional to partial molar entropy, the device aims to harvest the entropy of fusion and the associated configurational entropy change of bulk metal alloys. The envisioned energy harvester employs two identical alloy electrodes operating at two different temperatures across the solidus lines, with a thin electrolyte layer in between.

Results and discussion

Device design utilizing liquid to solid phase transition

The electrode thermopower is essentially the derivative of chemical potential with respect to temperature , which is directly proportional to the isobaric partial molar entropy . Thus, the thermogalvanic devices utilizing a combination of liquid and solid electrodes exploit the entropy differences upon melting phase transformation. For instance, typical liquid thermocells employ liquid electrodes on both electrodes (Fig. 1a). If the cold-side electrode undergoes solidification upon cooling the device below the melting point, the cold-side electrode involves an abrupt change in thermopower (Fig. 1b). The vastly different thermopower between the two electrodes, when coupled to a temperature gradient across the melting point, generates an increased voltage difference upon cooling. Identical principles can be employed upon heating solid electrodes. The detailed thermodynamic model is described in Supplementary Note 1 (ESI†).

Fig. 1 Solid-to-liquid phase transition-driven thermopower boosted thermogalvanic energy harvesters. (a) Schematic comparing the thermogalvanic voltages generated using liquid–liquid electrodes and liquid–solid electrodes. (b) Temperature–molar entropy diagrams showing the working principles behind thermogalvanics exploiting entropy of fusion. (c) Na–K phase diagrams showing the two targeted phase transition regimes including the eutectic composition (−12.6 °C eutectic temperature) and Na_2+xK alloys (6.9 °C solidus temperature).

To validate this working principle, we employ Na–K binary alloys as model systems (Fig. 1c). The Na–K alloy system provides a wide range of phases available near the room temperature, making the system ideal to explore the effect of solid-to-liquid phase transition.^21,22 We employ eutectic Na–K alloys (NaK-77, 77 wt% K and 23 wt% Na) and Na-rich Na_2+xK (28.3 wt% K and 71.7 wt% Na, melting point 45 °C) alloys to demonstrate the thermopower enhancement below room temperature. As the eutectic alloy gets cooled across the eutectic point, the alloy solidifies into a mixture of Na₂K and K alloys losing both vibrational entropy and configurational entropy from liquid mixtures. This results in notably decreased thermopower on the solidified electrode and thus involves a large thermopower difference and the generated voltage across the two electrodes. The electrodes based on liquid Na_2+xK alloys undergo continuous phase transition upon cooling, with pure Na nucleated from the alloy upon crossing the liquidus line. In this temperature regime between liquidus and solidus lines, Na exists as a mixture of liquid and solid state, with their proportion gradually decreasing until reaching the solidus temperature. On the other hand, K remains as a liquid below the liquidus temperature until it undergoes an instantaneous liquid-to-solid phase transition at the solidus temperature. The TGHs employing identical Na_2+xK electrodes on both sides but with either Na⁺ or K⁺ electrolytes thus work as a platform to compare the thermopower exploiting Na or K chemical potentials and verify the boosting mechanism described above.

Device performance upon temporal temperature changes

To verify the envisioned thermopower enhancement upon phase transformation, we first demonstrate the devices based on coin cells employing the liquid alloy electrode on one side (Fig. 2a). The designed thermogalvanic devices target temporal temperature changes and generate voltage by exploiting the thermopower difference between the two electrodes. The liquid Na–K alloys are incorporated into copper foam for morphological and electrical stability (Fig. S1, ESI†). Without the Cu foam, liquid alloys often resulted in poor electrical connection as they float above electrolytes due to low density or significantly increased charge transfer resistance in the electrochemical impedance spectroscopy (EIS) analysis (Fig. S2 and S3, ESI†). Carbon paper-based alloy composites were also tested, yet resulted in unstable impedance possibly due to carbon incorporation. The fabricated composite electrodes at the eutectic composition, along with K counter electrodes and K⁺ electrolytes are incorporated into coin cells for testing. When the coin cell is cooled from 25 °C to −5 °C at intervals of 10 °C, the observed voltage difference and the respective thermopower remains small; the average thermopowers are 0.14 mV K⁻¹, 0.3 mV K⁻¹ and 2.3 mV K⁻¹ for 25 to 15 °C, 15 to 5 °C and 5 to −5 °C intervals, respectively (Fig. 2b). In the interval from −5 to −15 °C across the euctectic point at −12.6 °C, however, the voltage increases sharply by 112 mV on average, achieving a high thermopower of 11.2 mV K⁻¹vs. K metal. This clear thermopower enhancement, which was repeated in a stable manner, is attributed to the increase in partial molar entropy upon phase transition.

Fig. 2 Melting phase transition-driven thermopower enhancement demonstrated among Na–K alloys. (a) Schematic of coin-cell type energy harvesters harvesting temporal temperature changes. One electrode undergoes phase transition over time and changes voltages higher than the other electrode, generating a thermogalvanic voltage. (b) Measured voltage changes and thermopower upon temporal temperature changes with the eutectic Na–K alloys on one electrode and a K counter electrode. Measured voltages and thermopower with the Na_2+xK working and K counter electrodes equipped with (c) K⁺ electrolytes and (d) Na⁺ electrolytes. Noticeable increase in thermopower is observed only when the temperature reaches below the solidus temperature with K⁺ electrolytes.

To further elucidate the effect of liquid-to-solid phase transition on the observed thermopower, we test coin cells with Na_2+xK electrodes and K counter electrodes. NaClO₄ and KPF₆ salts are used for Na⁺ and K⁺ electrolytes, respectively. Both electrolytes employ eutectic-forming ethylene carbonate (EC) and dimethyl carbonate (DMC) solvents in a 1 : 1 ratio with 10 vol% FEC additives for controlled electrode–electrolyte interaction and stable operation at low temperatures near 4 °C (Fig. S4, ESI†). As the temperature is decreased from 10 °C to 4 °C, a notable increase in voltage is observed for the coin cells containing K⁺ electrolytes (Fig. 2c). The observed thermopower reaches +5.8 mV K⁻¹vs. K metal. As the temperature gets cooled from 30 °C to 8 °C, the cell exhibits a thermopower of only +0.7 mV K⁻¹vs. K metal. This demonstrates that crossing the solidus line at 7 °C and inducing phase transition significantly increases the thermopower. Performing the identical experiments with Na⁺ electrolytes results in notably different thermopower. The observed thermopower upon cooling from 10 °C to below 4 °C is only approximately 0.3 mV K⁻¹vs. Na metal across the 10 °C to 2 °C temperature range (Fig. 2d), consistent with the device design in Fig. 1c. To further confirm the role of phase transition on the thermopower enhancement, we accurately controlled the temperature of the K⁺ electrolyte-containing coin cell by 1 °C from 9 °C to 6 °C. The observed voltage change is notably large from 7 °C to 6 °C, compared to the other temperature ranges (Fig. S5, ESI†). From the results measured with very small temperature changes near the phase transition temperature, it can be seen that the changes in variables such as ionic conductivity with temperature are negligible. Instead, the phase transition itself causes a significantly large increase in thermopower, dominating over other factors. These results match well with the hypotheses based on the phase diagram in Fig. 1c, confirming the phase transition-driven thermopower enhancement. The reported values are average values from experiments repeated multiple times. It is also noted that the enhanced thermopower demonstrated with several different combinations of electrode materials, experimental setup and device type indicates firm reproducibility of phase transition-induced thermopower enhancement.

Electrolyte optimization for the phase transition-driven thermopower

It is anticipated that melting electrodes undergo breakdown of the solid electrode–electrolyte interface (SEI) and potential overpotentials associated with the interfaces, especially during repeated phase transition. To accurately assess the phase transition-driven thermopower and minimize the effect of morphological instability at the SEI, we tested various Na⁺ and K⁺ electrolytes in TGHs. All carbonate-based electrolytes contained 10 vol% FEC to stabilize the liquid metal surface in terms of SEI (Fig. S6, ESI†).²³ K⁺ electrolytes based on ethers and KFSI salts are carefully tested to ensure that Na⁺ ions do not dissolve from liquid Na_xK electrodes and mix into the electrolytes (Fig. S7, ESI†). All electrolytes are tested for freezing behavior and liquid metal stability prior to testing the low temperature regimes (Fig. S8 and S9, ESI†).

Table 1 summarizes the electrodes and electrolyte combinations tested. The control experiments employing K and Na₂K electrodes with carbonate electrolytes exhibit 0.5 mV K⁻¹ and 1 mV K⁻¹ without phase transition, respectively. With Na₂K and NaK-77 electrodes that encompass a range of phases in the testing temperature regime, the measured thermopower varies from 0.7 mV K⁻¹ to 3.8 mV K⁻¹ as the electrolyte is varied from carbonate-based electrolyte to localized high-concentration electrolytes (LHCE) based on monoglyme ether (DME) and KFSI salts.²⁴ K⁺ electrolytes employing 2 M KFSI/50 mM KNO₃ salts in diglyme ether (DEGDME) provide a high average thermopower of 18.4 mV K⁻¹ (Fig. S10a, ESI†) when operated as a TGH across the spatial temperature gradient between 4 °C and 8 °C, due to the high stability of diglyme ether against reactive metal electrodes and high salt concentration that may replenish K⁺ ions upon repeated SEI formation.²⁵ Since liquid metal electrodes exhibit higher reactivity with electrolytes than solid metal electrodes,²⁶ protecting the electrode surfaces against morphological instability with 50 mM KNO₃ additives to further stiffen the SEI layers proves effective.^27,28 The TGHs employing the electrolyte successfully demonstrate a higher thermopower in the phase transition regime than that in liquid phase regime (10.8 mV K⁻¹, 8–12 °C) for K (Fig. S10b, ESI†). However, the short circuit current generated across the 4–8 °C temperature range is 328.5 nA/cm² on average, slightly lower than 333.6 nA cm⁻² across the 8–12 °C liquid temperature range (Fig. S11, ESI†). This may be attributed to the lowered K⁺ ionic conductivities in the electrolyte at low temperature. The electrochemical impedance spectroscopy (EIS) carried out at 25, 5 and −15 °C shows significantly increasing charge transfer impedance upon cooling (Fig. S12a, ESI†). The ionic conductivity slightly decreases as it passes through the eutectic point of NaK-77 at −12.6 °C and undergoes a solid phase transition, but the change is not significant (Fig. S12b, ESI†).

Table 1 The maximum thermopower obtained in various electrodes and electrolyte combinations

Type	Electrode 1	Electrode 2	Salt	Concentration	Solvent	Additive	Max. α (mV K^−1)	T Range (°C)	Note
Spatial T Gradient	K	K	KPF6	0.8 M	EC:DEC	FEC	0.5	30–60	K control experiment
Spatial T Gradient	Na2K	Na2K	NaClO4	1 M	EC:DEC	FEC	1.0	3–12	Na control experiment
Temporal	Na2K	Na2K	KFSI	0.5 M	EC:DMC	FEC	0.7	4–20	Solidus temperature (6.9 °C)/eutectic point (−12.6 °C)
Temporal	K	Na2K	KFSI	1 M	DME	n/a	11.2	−15 t0 25	Eutectic point (−12.6 °C)
Spatial T Gradient	Na2+xK	Na2+xK	KFSI	2 M	DEGDME	KNO3	18.4	4–8	Solidus temperature (6.9 °C)
Spatial T Gradient	Na2+xK	Na2+xK	KPF6/KFSI	0.2 M/1.8 M	DEGDME	KNO3	3.5	4–8	Solidus temperature (6.9 °C)
Spatial T Gradient	Na2+xK	Na2+xK	KPF6/KFSI	0.2 M/1.8 M	DEGDME/TTE5	KNO3	6.5	4–8	Solidus temperature (6.9 °C)
Spatial T Gradient	Na2+xK	Na2+xK	KPF6	1 M	DEGDME/TTE5	LiNO3	26.1	4–14	Solidus temperature (6.9 °C)

---

To further optimize the electrolyte in terms of both thermopower and impedance, we carefully engineered the electrolyte composition (Supplementary Note 2, ESI†). Mixing KPF₆ salt to lower the impedance results in a sharply decreased thermopower from 18.4 mV K⁻¹ to 3.5 mV K⁻¹ (Table 1). Mixing 5% 1,1,2,2-tetrafluoroethyl-2,2,3,3-tetrafluoropropylether (TTE) as the co-solvent results in a stabilized SEI and improves the thermopower when employed with KPF₆ salts. The optimized electrolyte composition is chosen to be 1 M KPF₆ dissolved in DEGDME with 5% TTE and 20 mM LiNO₃. Similar electrolyte composition with DME solvent has been reported for effective low-temperature K-ion batteries.²⁹ EIS analysis was also conducted on the optimized electrolyte (Fig. S13a, ESI†). The optimized electrolyte did not exhibit a significant decrease in charge transfer impedance even when the temperature was lowered to 5 °C, unlike the 2 M KFSI DEGDME electrolyte. Additionally, the ionic conductivity did not significantly decrease when passing through the solidus temperature of 6.9 °C for Na_2+x (Fig. S13b, ESI†).

Based on the electrolyte optimization, the TGH device is fabricated with two identical Na_2+xK alloy-based electrodes and the KPF₆ in DEGDME electrolyte (Fig. 3a). When one side is repeatedly heated to 14 °C from the base temperature of 4 °C (ΔT = 10 K), the device generates average voltage increases of 261 mV, equivalent to the average thermopower of 26.1 mV K⁻¹ (Fig. 3b). We measured the short circuit current under the same temperature conditions of 4–14 °C (Fig. 3c). The thermal current generated under an identical setup reaches up to 72 μA cm⁻² under the ΔT = 10 K condition. The thermal current shows a gradual decrease. This decline is attributed to the electrolyte evaporating over time. The device produces stable current output for 14 repeated temperature cycles during the continued experiments of 5 hours without major signs of degradation. The stability of the cells is further confirmed for continuous experiments for 25 hours and 45 repeated thermal cycles (Fig. S14, ESI†). Furthermore, the power output was measured by connecting external loads to the thermogalvanic device and measuring the currents across the external loads (Fig. S15, ESI†). As external resistance increases beyond the device's internal resistance, the measured current continuously decreases, while it increases below the internal resistance, reaching a short circuit current of 60 μA. A maximum power output of 4.0 μW cm⁻² is achieved when the external load is similar to the internal resistance.

Fig. 3 Optimized performance of thermogalvanic cells exploiting phase-changing Na_2+xK electrodes. (a) Schematic of thermogalvanic devices with identical solid and liquid alloy electrodes targeting spatial temperature gradients. (b) The generated voltages and measured average thermopower in the 4–14 °C temperature range across the solidus temperature. (c) The generated short circuit current in an identical 4–14 °C temperature range. The device has areal dimensions of 1 × 1 cm² and a one-side electrode thickness of 0.64 cm.

When compared to previously reported thermogalvanic devices or thermally regenerative electrochemical cells (TREC), the thermopowers demonstrated in this work notably include a high thermopower of 26.1 mV K⁻¹ among TGHs (Fig. 4). The comparison was made within thermogalvanic devices as categorized by a recent review.¹³ Most thermogalvanic thermopowers reported near room temperature so far remained around the 3–6 mV K⁻¹ range,³⁰ often employing the redox activities of solvated ferricyanides,^17,31–34 iodide ions,³⁵ iron ions,³⁶ protonated cobalt(II/III) sarcophagine³⁷ or quinhydrone.³⁸ The highest thermopower demonstrated among thermogalvanic devices so far is 9.9 mV K⁻¹ for the liquid to vapor phase transition utilizing acetylene-isopropyl alcohols.³⁹ Similar enhancements have also been reported for crystallization of Fe(CN)₆⁴⁻ in the liquid thermocell with a demonstrated thermopower of 4.2 mV K⁻¹.¹⁶ Compared to these results, we first note that utilizing melting phase transitions result in significantly boosted thermopower for small temperature ranges that exceeds those in the liquid regime with higher absolute entropies. Furthermore, engineering the interaction between the electrode–electrolyte interface during melting phase transition is crucial to evaluating the thermopower improvement during phase transition.

Fig. 4 Performance comparison of the demonstrated thermopower and the average working temperature for this work and previous literature. Exploiting the solid-to-liquid phase transition results in notably enhanced thermopower. The filled-in symbols show those tested with single-phase electrodes while the empty ones show those with phase-transition electrodes.

Experimental

Materials preparation

Copper foam (EQ-bccf-1mm, thickness of 1 mm) was purchased from MTI Korea. The Cu foam was cut into 15 mm diameter for coin-cell-based thermal energy harvesters and 10 × 10 mm squares for use in the optimized thermogalvanic cell with 1 MPa pneumatic pressing. Carbon cloth (WOS1009) and carbon paper (HCP030N) were used as substrates for comparison with Cu foam. Potassium and sodium metals were purchased from Sigma Aldrich and mixed in two different ratios, Na_2+xK (71.7 wt% Na) with a melting point of 45 °C and NaK-77. The NaK alloy and K were then infiltrated into the prepared Cu foams on a heating plate at 380 °C in an Ar-filled glove box and subsequently cooled to ambient temperature.

All solvents used, ethylene carbonate (EC), dimethyl carbonate (DMC), diethyl carbonate (DEC), polypropylene carbonate (PC), fluoroethylene carbonate (FEC), 1,1,2,2-tetrafluoroethyl-2,2,2-trifluoroethyl ether (TFTFE), dimethoxyethane (DME) and diethylene glycol dimethyl ether (DEGDME), were purchased from Sigma Aldrich and dried for 2 days with freshly activated molecular sieves (4 Å). Sodium perchlorate (NaClO₄) and potassium hexafluorophosphate (KPF₆) were purchased from Sigma Aldrich and potassium bis(fluorosulfonyl)imide (KFSI) was purchased from Tokyo Chemical Industry. All salts were dried before use at 80 °C in a vacuum oven for 24 hours.

Carbonate-based electrolytes were prepared by mixing EC with DMC or DEC in a 1 : 1 ratio and adding salts at concentrations of 0.5 M, 0.8 M and 1 M. Ether-based electrolytes were prepared by adding KPF₆ or KFSI to DME or DEGDME at concentrations of 1 M and 2 M. Additionally, a localized high-concentration electrolyte was made by mixing KFSI, DME and TFEFE in a 1 : 2 : 2 ratio.

Device fabrication

For preparing coin cell-based thermal energy harvesters, NaK-77 infiltrated Cu foam was squeezed with a finger to remove excess NaK-77 and prevent escaping of NaK-77 from Cu foam after coin-cell crimping. Cu foam with a NaK-77 alloy and K were assembled into CR2032 coin cells with a 25-μm thick Celgard PE separator in a glove box in an environment with H₂O < 0.1 and O₂ < 0.1 ppm. 1 M KFSI DME electrolyte was used. The cell was crimped using a pneumatic press at 7 MPa.

For the thermogalvanic cell, five Cu foams with Na_2+xK alloy were placed into two copper blocks (30 × 30 × 6.4 mm with 10 × 10 × 6.4 mm hole for Cu foams with Na_2+xK alloy) respectively. The holes of the blocks were covered with Cu foil used as a lead wire and PET tape for the battery was used to seal and fix the Cu foil and PE blocks. A 25-μm thick Celgard PE separator was soaked with 1 M KPF₆ DEGDME electrolyte and put between the block, and the blocks were sealed with the PET tape. Two polyethylene-sheathed thermocouples were inserted into each block through small holes with penetrating Cu foams to measure the temperature of the Na_2+xK alloy.

Electrochemical characterization

The assembled cells were tested using a Gamry 1000E potentiostat. The open circuit potential was measured with a potentiostat with specific temperature profiles. The temperatures were measured by thermocouples and a PT100 RTD sensor and controlled by Arduino and thermoelectric modules via feedback-PID control. Electrochemical impedance spectroscopy (EIS) was used to measure impedance dependence according to temperature change. Short circuit currents were measured using Keysight 34465A.

Conclusions

In this work, we show that the entropy changes associated with solid-to-liquid phase transition can be utilized to boost the thermopower near room temperature. The fabricated thermogalvanic device generates a thermopower of 26.1 mV K⁻¹ across the solidus line, demonstrating the phase transition-driven thermopower enhancement. We also report that engineering the kinetic overpotentials and interaction among liquid electrolytes and electrodes is critical to stabilizing electrochemical stability during phase transition and thus achieving high thermopower. The strategy is generalizable to other alloy systems with various eutectic points and solidus temperatures, as well as to potential energy storage mechanisms such as thermally chargeable capacitors with active liquid surfaces. With the enhanced thermopower across the melting point, this work signals a design strategy for thermogalvanic devices that maximize energy harvesting performances at specifically targeted temperature ranges. This work has broad implications in low-grade thermal energy recovery. By attaching TGHs to specific heat sources such as electronic devices, human bodies, car exhaust systems, or heat pipes, energy can be harvested from temperature differences between the ambient environment and the heat source. Furthermore, designing tandem TGHs connected in series, with electrodes of varying melting points for each cell, optimizes the harvesting performance to capture larger temperature differentials.

Author contributions

D. Shin (methodology: lead; investigation: lead; visualization: lead; writing – original draft: lead; writing – review & editing: lead). K. Ryu (methodology: lead; investigation: lead; visualization: lead; writing – review & editing: lead). D. Kim (methodology: supporting; investigation: supporting). E. Choi (methodology: supporting; investigation: supporting). S. Chae (methodology: supporting; investigation: supporting). Y. Lee (methodology: supporting; investigation: supporting). Y. Kang (funding acquisition: supporting; project administration: supporting; supervision: supporting; writing – review & editing: supporting). S. Kim (conceptualization: lead; funding acquisition: lead; project administration: lead; supervision: lead; writing – original draft: lead; writing – review & editing: lead). W. Choi (conceptualization: lead; funding acquisition: lead; project administration: lead; supervision: lead; writing – original draft: lead; writing – review & editing: lead).

Data availability

The authors confirm that the data supporting the findings of this study are available within the article or its ESI.†

Conflicts of interest

The authors declare no competing interests.

Acknowledgements

This work was supported by the Samsung Science & Technology Foundation under Project Number SRFC-MA-1902-07, and a National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science and ICT) (No. 2023R1A2C2006407, 2020R1A5A1018153, RS-2023-00212932, RS-2023-00212932, and NRF-2021M2D2A2076378). S. K. appreciates Prof. Ju Li (MIT) for the helpful discussions.

References

1. A. J. Simon, How to Read an LLNL Energy Flow Chart (Sankey Diagram), Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States), 2016.
2. H. Jouhara, N. Khordehgah, S. Almahmoud, B. Delpech, A. Chauhan and S. A. Tassou, Therm. Sci. Eng. Prog., 2018, 6, 268–289.
3. B. Burrows, J. Electrochem. Soc., 1976, 123, 154.
4. M. Li, M. Hong, M. Dargusch, J. Zou and Z.-G. Chen, Trends Chem., 2021, 3, 561–574.
5. T. Li, X. Zhang, S. D. Lacey, R. Mi, X. Zhao, F. Jiang, J. Song, Z. Liu, G. Chen, J. Dai, Y. Yao, S. Das, R. Yang, R. M. Briber and L. Hu, Nat. Mater., 2019, 18, 608–613.
6. M. A. Rahman and M. Z. Saghir, Int. J. Heat Mass Transfer, 2014, 73, 693–705.
7. J. Wang, S.-P. Feng, Y. Yang, N. Y. Hau, M. Munro, E. Ferreira-Yang and G. Chen, Nano Lett., 2015, 15, 5784–5790.
8. F. Garmroudi, M. Parzer, A. Riss, C. Bourgès, S. Khmelevskyi, T. Mori, E. Bauer and A. Pustogow, Sci. Adv., 2023, 9, eadj1611.
9. Y. Lu, Y. Zhou, W. Wang, M. Hu, X. Huang, D. Mao, S. Huang, L. Xie, P. Lin, B. Jiang, B. Zhu, J. Feng, J. Shi, Q. Lou, Y. Huang, J. Yang, J. Li, G. Li and J. He, Nat. Nanotechnol., 2023, 18, 1281–1288.
10. P. Zhao, W. Xue, Y. Zhang, S. Zhi, X. Ma, J. Qiu, T. Zhang, S. Ye, H. Mu, J. Cheng, X. Wang, S. Hou, L. Zhao, G. Xie, F. Cao, X. Liu, J. Mao, Y. Fu, Y. Wang and Q. Zhang, Nature, 2024, 631, 777–782.
11. C. B. Vining, Nat. Mater., 2009, 8, 83–85.
12. H. J. Kim, Y. Park, Y. Kwon, J. Shin, Y.-H. Kim, H.-S. Ahn, R. Yazami and J. W. Choi, Energy Environ. Sci., 2020, 13, 286–296.
13. Y. Liu, M. Cui, W. Ling, L. Cheng, H. Lei, W. Li and Y. Huang, Energy Environ. Sci., 2022, 15, 3670–3687.
14. K. Xu, Chem. Rev., 2004, 104, 4303–4418.
15. P. A. Linford, MS thesis, Massachusetts Institute of Technology, 2017.
16. B. Yu, J. Duan, H. Cong, W. Xie, R. Liu, X. Zhuang, H. Wang, B. Qi, M. Xu, Z. L. Wang and J. Zhou, Science, 2020, 370, 342–346.
17. J. Duan, G. Feng, B. Yu, J. Li, M. Chen, P. Yang, J. Feng, K. Liu and J. Zhou, Nat. Commun., 2018, 9, 5146.
18. Y. Han, J. Zhang, R. Hu and D. Xu, Sci. Adv., 2022, 8, eabl5318.
19. C.-G. Han, X. Qian, Q. Li, B. Deng, Y. Zhu, Z. Han, W. Zhang, W. Wang, S.-P. Feng, G. Chen and W. Liu, Science, 2020, 368, 1091–1098.
20. Z. Li, Y. Xu, L. Wu, J. Cui, H. Dou and X. Zhang, Nat. Commun., 2023, 14, 6816.
21. S. Zhang, Y. Liu, Q. Fan, C. Zhang, T. Zhou, K. Kalantar-Zadeh and Z. Guo, Energy Environ. Sci., 2021, 14, 4177–4202.
22. L. Zhang, X. Xia, Y. Zhong, D. Xie, S. Liu, X. Wang and J. Tu, Adv. Mater., 2018, 30, 1804011.
23. T. Hou, G. Yang, N. N. Rajput, J. Self, S.-W. Park, J. Nanda and K. A. Persson, Nano Energy, 2019, 64, 103881.
24. X. Cao, H. Jia, W. Xu and J.-G. Zhang, J. Electrochem. Soc., 2021, 168, 010522.
25. L. Li, S. Zhao, Z. Hu, S.-L. Chou and J. Chen, Chem. Sci., 2021, 12, 2345–2356.
26. T. Hosaka, T. Matsuyama, K. Kubota, R. Tatara and S. Komaba, J. Mater. Chem. A, 2020, 8, 23766–23771.
27. W. Jia, C. Fan, L. Wang, Q. Wang, M. Zhao, A. Zhou and J. Li, ACS Appl. Mater. Interfaces, 2016, 8, 15399–15405.
28. H. Wang, J. Dong, Q. Guo, W. Xu, H. Zhang, K. C. Lau, Y. Wei, J. Hu, D. Zhai and F. Kang, Energy Storage Mater., 2021, 42, 526–532.
29. J. Chen, D. Yu, Q. Zhu, X. Liu, J. Wang, W. Chen, R. Ji, K. Qiu, L. Guo and H. Wang, Adv. Mater., 2022, 34, 2205678.
30. J. Duan, B. Yu, L. Huang, B. Hu, M. Xu, G. Feng and J. Zhou, Joule, 2021, 5, 768–779.
31. S. W. Lee, Y. Yang, H.-W. Lee, H. Ghasemi, D. Kraemer, G. Chen and Y. Cui, Nat. Commun., 2014, 5, 3942.
32. X. Wang, Y.-T. Huang, C. Liu, K. Mu, K. H. Li, S. Wang, Y. Yang, L. Wang, C.-H. Su and S.-P. Feng, Nat. Commun., 2019, 10, 4151.
33. H. Zhang, D. G. Lek, S. Huang, Y. M. Lee and Q. Wang, Adv. Mater., 2022, 34, 2202266.
34. D. Zhang, Y. Mao, F. Ye, Q. Li, P. Bai, W. He and R. Ma, Energy Environ. Sci., 2022, 15, 2974–2982.
35. H. Wang, X. Zhuang, W. Xie, H. Jin, R. Liu, B. Yu, J. Duan, L. Huang and J. Zhou, Cell Rep. Phys. Sci., 2022, 3, 100737.
36. Y. Liu, Q. Zhang, G. Omololu Odunmbaku, Y. He, Y. Zheng, S. Chen, Y. Zhou, J. Li, M. Li and K. Sun, J. Mater. Chem. A, 2022, 10, 19690–19698.
37. K. Laws, M. A. Buckingham, M. Farleigh, M. Ma and L. Aldous, Chem. Commun., 2023, 59, 2323–2326.
38. B. Guo, Y. Hoshino, F. Gao, K. Hayashi, Y. Miura, N. Kimizuka and T. Yamada, J. Am. Chem. Soc., 2020, 142, 17318–17322.
39. H. Zhou and P. Liu, ACS Appl. Energy Mater., 2018, 1, 1424–1428.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ee01642d

‡ These authors contributed equally to this work.

---