Abstract
Thermoelectric materials capable of operating at 600–900 °C are attractive for recovering industrial and nuclear waste heat toward a carbon-neutral society. Here, we report the structure-guided enhancement of high-temperature thermoelectric performance in natural weathered biotite (WB) through molten-salt treatment using a KCl–CaCl2 mixed salt (molar ratio 3:1) combined with systematic process optimization. The effects of WB particle size (<45 and 45–75 μm), salt addition ratio (1/100–1/1 by mass relative to WB), and compaction pressure (150–200 MPa) on crystal structure, densification behavior, and thermoelectric transport properties were systematically examined. Structural and compositional analyses revealed progressive incorporation of K and Ca with increasing salt addition while largely preserving the layered WB framework, together with additional diffraction features indicating reaction-induced structural modifications. Among the conditions investigated, samples prepared from WB with an intermediate particle size of 45–75 μm exhibited relatively high packing densities (>70%) at low salt addition levels and showed enhanced electrical conductivity, whereas excessive particle refinement (<45 μm) suppressed electrical transport, indicating that the transport response is sensitive to particle-size-dependent packing and processing-induced microstructural variation. High-temperature measurements further revealed exceptionally large Seebeck coefficients of up to ∼1.9 × 105 μV K–1, indicating that nonelectronic contributions may be involved in the observed thermovoltage. At present, the origin of this anomalously large response has not been conclusively established and requires further dedicated investigation. As a result of the cooperative optimization of particle size, salt addition, and compaction pressure, a maximum thermoelectric figure of merit of ZT = 0.29 at 667 °C was achieved, representing a marked improvement in this material system. These results demonstrate that microstructural control and densification strongly influence the measured thermoelectric response in layered silicate minerals and provide useful process–structure–property insights for natural mineral-derived high-temperature thermoelectric materials.
1. Introduction
Thermoelectric energy conversion is a promising approach for realizing a carbon-neutral society through the efficient utilization of waste heat. In particular, thermoelectric materials capable of operating at 600–900 °C are highly desirable for harvesting industrial and nuclear waste heat. This temperature range is of particular importance because substantial amounts of waste heat above 500 °C are emitted from energy-intensive sectors, including the steel industry in Europe, underscoring the technological need for thermoelectric materials capable of high-temperature operation. Because thermoelectric materials directly convert a temperature gradient into electrical power and operate without moving parts, they offer high reliability, silent operation, and low maintenance. These features make them attractive for next-generation energy recovery technologies. Conventional thermoelectric materials, including Bi2Te3-, PbTe-, and SiGe-based systems, exhibit high performance with large dimensionless figures of merit (ZT). − However, their practical application is limited by the use of scarce or toxic elements and high material costs. Consequently, considerable effort has been devoted to developing alternative thermoelectric materials composed of abundant, nontoxic, and low-cost elements.
Following the 2011 Great East Japan Earthquake and the subsequent Fukushima Daiichi Nuclear Power Plant accident, our research group has been engaged in environmental recovery studies. , In this context, we focused on weathered biotite (WB), a naturally abundant layered silicate mineral widely distributed in Fukushima Prefecture, and developed molten-salt heat-treatment methods using NaCl–CaCl2 and KCl–CaCl2 systems. These approaches enabled the synthesis of two classes of WB-derived materials: a structure-preserving type that retains the layered framework of WB and a structure-transformed type in which the original layered structure is converted into a different crystalline phase. Importantly, these materials exhibited thermoelectric activity at 700–850 °C, a temperature range highly relevant to industrial waste-heat recovery, and our previous study reported a maximum ZT of 0.015 for this material system. These findings highlighted the potential of WB-derived materials as thermoelectric candidates for practical high-temperature operation.
In addition to phase optimization, control of microstructural parameters such as particle size and densification behavior is essential for further improving thermoelectric performance. , Particle size refinement and nanostructuring can enhance phonon scattering and thereby reduce lattice thermal conductivity. , However, excessive particle size reduction also increases grain-boundary density, which can significantly deteriorate electrical conductivity. Therefore, identifying an optimal particle size is critical for maximizing thermoelectric performance. Similar structure–transport correlations have been reported in conventional thermoelectric systems such as Bi0.3Sb1.7Te3 and half-Heusler alloys. ,
Furthermore, compaction pressure during pellet formation and the mixing ratio between WB and molten salt are also expected to strongly influence the densification and transport pathways of the resulting materials. , Consequently, the cooperative optimization of these processing parameters is essential for achieving enhanced thermoelectric performance.
Based on this background, the present study aims to enhance the thermoelectric performance of WB-based materials by systematically investigating the effects of three key processing parameters: the particle size of WB powders (45–75 μm and <45 μm), the addition ratio of KCl–CaCl2 mixed salt (molar ratio 3:1; mass ratio 1/100–1/1 relative to WB), and the compaction pressure during pellet formation (150–200 MPa). By correlating these factors with crystal structure, densification behavior, and thermoelectric transport properties, this work provides process–structure–transport insight into naturally derived layered silicate materials and establishes a basis for further evaluation of their high-temperature thermoelectric response.
2. Materials and Methods
2.1. Sample Preparation (Grinding, Sieving, and Particle Size Distribution)
Weathered biotite (WB) used in this study was obtained as a natural mineral collected in Fukushima Prefecture, Japan. To control particle size, the raw WB was first ground using a vibratory ball mill (MM400, Retsch). Grinding was performed at 15 Hz for 30 min in a 50 mL stainless-steel jar containing a single 20 mm stainless-steel ball. The resulting powders were subsequently classified using a vibratory sieve shaker (AS200, Retsch) equipped with stainless-steel meshes of 150, 100, 75, and 45 μm, arranged in descending order. Sieving was conducted in interval mode with an amplitude of 1.00 mm for a total duration of 50 min. Using this procedure, two operational particle size fractions were obtained: a fine-particle fraction (<45 μm) and an intermediate-particle fraction (45–75 μm). Particle size distributions were evaluated by laser diffraction using a particle size analyzer (Partica mini LA-350, HORIBA) with a wet dispersion method. All particle size data are reported on a volume basis. The resulting particle size distributions are shown in Figure , and the corresponding characteristic particle sizes are summarized in Table . Grinding with the MM400 produced a broad particle size distribution; subsequent sieving was used to obtain two operational particle size fractions for further comparison. For comparison, the particle size distribution of WB obtained in previous work using hydraulic classification is also included in Figure
1.
Volume-based particle size distributions of weathered biotite (WB) powders prepared under different grinding and classification conditions, as measured by laser diffraction with wet dispersion. The black, blue, green, and purple curves represent WB prepared by hydraulic classification in previous work, MM400-ground WB without sieving, and the <45 μm and 45–75 μm sieve fractions after MM400 grinding, respectively.
1. Volume-Based Particle Size Metrics (D10, D50, and D90) of Weathered Biotite (WB) Powders Prepared by Grinding and Sieving, as Evaluated by Laser Diffraction with Wet Dispersion.
| sample | D10 (μm) | D50 (μm) | D90 (μm) | span |
|---|---|---|---|---|
| (a) 45–75 μm (sieved) | 2.89 | 8.73 | 23.71 | 2.38 |
| (b) <45 μm (sieved) | 4.26 | 16.7 | 43.56 | 2.35 |
| before sieving (MM400 ground) | 4.29 | 13.80 | 48.88 | 3.23 |
| hydraulic classification (previous work) | 0.38 | 2.81 | 8.39 | 2.85 |
Span = (D90 – D10)/D50.
2.2. Molten-Salt Treatment and Salt Addition Ratio
A KCl–CaCl2 mixed salt with a molar ratio of 3:1 was selected for the present work based on our previous study, in which this composition gave the highest thermoelectric performance in the material system (ZT = 0.015 at 747 °C). To systematically investigate the effects of salt content on reaction behavior and thermoelectric performance, the amount of molten salt was varied over five mass ratios relative to WB: 1/100, 1/50, 1/20, 1/10, and 1/1.
For sample preparation, predetermined amounts of WB powder and the mixed salt were homogeneously blended and placed in a platinum crucible, which was then introduced into a ring-type electric furnace. The samples were heated at a rate of 15 °C min–1 up to 700 °C, held at this temperature for 2 h, and subsequently cooled naturally inside the furnace. After heat treatment, the obtained reaction products were washed with ultrapure water to remove residual salt components. Specifically, the samples were dispersed in 50 mL centrifuge tubes, stirred at 80 °C for 30 min using a servo mixer, and then centrifuged at 3000 rpm for 30 min. This washing–centrifugation process was repeated multiple times to remove residual salt. The washed samples were then dried in an oven at 80 °C.
The dried powders were subsequently subjected to chemical composition and crystal structure analyses, while separate portions were used for pellet fabrication followed by thermoelectric property measurements. Chemical composition and crystal structure were analyzed by X-ray fluorescence (XRF) and X-ray diffraction (XRD), respectively, while thermoelectric transport properties were evaluated after pellet fabrication and sintering, as described in Sections and 2.4.
2.3. Compaction and Sintering
After washing and drying, the obtained powder samples were loaded into a cylindrical die with a diameter of 20 mm and uniaxially compacted under controlled pressures in the range of 150–200 MPa. The compaction pressure was varied to control the packing density of the green compacts and to examine its effect on thermoelectric transport. The green compacts were subsequently sintered in an electric furnace. The temperature was increased from room temperature to 900 °C at a constant heating rate of 100 °C h–1, held at 900 °C for 5 h, and then furnace-cooled to room temperature. After sintering, the bulk density of each sample was measured using a displacement-based method (GeoPyc, Micromeritics), while the true density was determined by helium gas pycnometry (AccuPyc, Micromeritics). The packing ratio was calculated as the ratio of bulk density to true density. These parameters were used to quantitatively evaluate the effects of particle size, molten-salt addition ratio, and compaction pressure on the densification behavior of the sintered materials.
2.4. Thermoelectric Property Measurements
The sintered samples were polished into rectangular bars with dimensions of 1 mm in thickness, 5 mm in width, and 15 mm in length. Electrical conductivity and Seebeck coefficient were measured sequentially on the same specimen using a thermoelectric measurement system (OZMA-1-S1, Ozawa Science Co., Ltd.). Thermal diffusivity and specific heat capacity were evaluated using a laser flash apparatus (LFA467, Netzsch Co., Ltd.). To assess specimen-to-specimen reproducibility within the same sintered body, additional electrical-conductivity measurements were performed for representative salt-to-WB mass ratios of 1/100, 1/50, and 1/1 using independently cut specimens measured under the same conditions. The Seebeck coefficient was evaluated after stabilization of the thermoelectric voltage under each temperature condition. For the LFA measurements, samples with dimensions of 10 mm × 10 mm were prepared, and both surfaces were coated with black carbon to improve measurement accuracy. The specific heat capacity of each sample was determined using molybdenum as a reference material. All thermophysical measurements were conducted over a temperature range of 650–850 °C. The thermal conductivity (κ) was calculated from the measured thermal diffusivity, specific heat capacity, and bulk density of the sintered samples, where the bulk density was obtained using GeoPyc. The dimensionless thermoelectric figure of merit, ZT, was calculated as ZT = S 2σT/κ, where S, σ, T, and κare the Seebeck coefficient, electrical conductivity, absolute temperature, and thermal conductivity, respectively. In the present study, the calculated ZT values are used for comparative evaluation within the WB-derived material system. Because the anomalously large Seebeck response may include nonelectronic contributions, these values should not be interpreted as rigorously established intrinsic electronic thermoelectric figures of merit.
3. Result and Discussions
3.1. Compositional and Phase Evolution with Varying Molten Salt Addition
The compositional and structural evolution of WB-derived materials treated with KCl–CaCl2 (molar ratio 3:1) was investigated as a function of salt addition amount. The salt content was systematically varied from 1/100 to 1/1 by weight relative to WB, and the resulting samples were analyzed by X-ray fluorescence (XRF) and X-ray diffraction (XRD).
Figure shows the K and Ca contents of the WB-derived materials as a function of KCl–CaCl2 addition ratio. The corresponding XRF spectra are provided in Figure S1 of the Supporting Information. For both particle-size fractions, the contents of K and Ca increased monotonically with increasing salt addition, confirming the effective incorporation of molten-salt-derived elements into the WB-derived structure. In contrast, no pronounced difference was observed between the <45 μm and 45–75 μm fractions, indicating that particle size has only a minor influence on the overall uptake of K and Ca under the present processing conditions. These results indicate that molten-salt treatment primarily causes compositional modification associated with K and Ca incorporation, whereas particle size mainly affects the subsequent structural/densification response rather than the overall elemental uptake. At the present stage, the thermoelectric enhancement is not directly assigned to any single newly formed phase.
2.
K and Ca contents of WB-derived materials as a function of KCl–CaCl2 addition ratio for the 45–75 μm sieve fraction: (a) K content and (b) Ca content. The untreated WB sample is included as a reference.
The XRD patterns (Figure ) reveal that the characteristic diffraction peaks associated with the layered structure of weathered biotite are largely preserved after molten-salt treatment, indicating that the fundamental WB-derived framework is retained to a large extent. At the same time, additional diffraction features appear over the salt addition range of 1/100–1/1, indicating reaction-induced structural modifications during molten-salt treatment. Because the present data do not allow rigorous phase identification or quantification of these additional features, the discussion is limited to qualitative structural evolution. No pronounced particle-size-dependent differences in the XRD patterns were detected, implying that molten-salt treatment modifies the chemical/structural state while maintaining the overall layered framework irrespective of the initial particle-size fraction.
3.
XRD patterns of WB-derived materials treated with KCl–CaCl2 (3:1) at different salt addition ratios for the (a) 45–75 μm and (b) <45 μm sieve fractions, together with an enlarged view of the selected diffraction region for the 45–75 μm fraction. The WB-derived layered framework is largely retained after molten-salt treatment, while additional salt-addition-dependent diffraction features appear in the enlarged region, indicating reaction-induced structural modifications. Arrows highlight representative additional diffraction features. Exact phase identification of these features remains unresolved.
It should be noted that the sieve-classified fractions and the particle size metrics obtained by laser diffraction do not exhibit a simple monotonic correspondence. As summarized in Table , the two fractions show comparable distribution widths (span ≈ 2.3–2.4), whereas the apparent D50 values do not directly follow the nominal sieve ranges. This discrepancy is reasonably attributed to agglomeration of finely ground WB particles and the anisotropic, plate-like morphology of the clay mineral, which influence both sieve passage behavior and the equivalent spherical diameter obtained by laser diffraction. Therefore, in the present study, the <45 μm and 45–75 μm samples are regarded as operationally defined sieve fractions with different packing and transport characteristics rather than as idealized monodisperse particle populations.
3.2. Effect of Molten Salt Addition and Particle Size on Electrical Conductivity
To evaluate the effect of KCl–CaCl2 (molar ratio 3:1) addition on electrical transport, the electrical conductivity (σ) was measured for WB-derived samples prepared with salt-to-WB mass ratios of 1/100, 1/50, 1/20, 1/10, and 1/1 (Figure ). Measurements were performed for two particle-size ranges obtained by milling and sieving: <45 μm and 45–75 μm.
4.
Temperature dependence of the electrical conductivity (σ) of WB-derived materials treated with KCl–CaCl2 (3:1) at different salt addition ratios: (a) 45–75 μm sieve fraction, (b) <45 μm sieve fraction, and (c) comparison between the two sieve fractions at the same salt addition ratio, together with the previously reported data.
As shown in Figure a, the 45–75 μm samples exhibited the highest electrical conductivity at the lowest salt addition ratio of 1/100. Increasing the salt content resulted in a gradual decrease in electrical conductivity, indicating that excessive molten salt addition does not improve electrical transport in this particle-size range. This trend indicates that the electrical-transport response depends nonmonotonically on salt addition in the intermediate-particle-size fraction. Excess salt addition may alter the packing behavior and processing-induced microstructural state in a manner unfavorable to the measured conductivity.
In contrast, the <45 μm samples (Figure b) showed the highest electrical conductivity at an intermediate salt addition ratio of 1/50, while no clear enhancement was observed for the other compositions, including the 1/100 condition. This behavior indicates that the response of electrical conductivity to salt addition is strongly particle-size dependent and reflects differences in reaction kinetics and microstructural evolution. In the fine-particle regime, an intermediate salt addition ratio produced the highest conductivity, indicating that the transport response depends on the combined effects of particle refinement, packing behavior, and salt-assisted structural modification.
To isolate the effect of particle size, the electrical conductivities of the two particle-size groups were compared at the same salt addition ratio of 1/1, as shown in Figure c. Under identical compositional conditions, the 45–75 μm samples exhibited approximately 1.5× higher electrical conductivity than the <45 μm samples, indicating a pronounced effect of sieve-classified powder fraction on conduction pathway formation. In view of the particle-size analysis, this difference is best interpreted not as a simple effect of nominal sieve size, but as a consequence of different agglomeration states, packing behavior, and interparticle connectivity generated by the two powder fractions.
To quantitatively support these trends, the bulk density, true density, and packing ratio of the sintered samples were evaluated, and the results are summarized in Table . For the 45–75 μm samples, particularly at salt addition ratios of 1/100 and 1/50, highly densified bodies with packing ratios exceeding 70% were obtained. The higher packing ratios observed for the 45–75 μm samples are consistent with the improved electrical conductivity measured for this fraction. In contrast, the <45 μm samples exhibited lower electrical conductivity even when relatively high packing ratios were achieved. This result indicates that densification alone is insufficient to account for the observed transport behavior in the fine-particle regime. Rather, the conductivity appears to depend on a combination of sieve-classified powder fraction, packing behavior, and processing-induced microstructural variation. Accordingly, the present results are interpreted as a process–structure–property correlation, while the detailed microscopic transport mechanism remains to be clarified.
2. Bulk Density, True Density, and Packing Ratio of Sintered WB-Derived Samples Prepared with Different Particle-Size Fractions and KCl–CaCl2 (3:1) Addition Ratios.
| particle size | salt addition ratio (mass ratio to WB) | bulk density (g/cm3) | true density (g/cm3) | packing ratio (%) |
|---|---|---|---|---|
| <45 μm | 1/100 | 1.71 | 3.06 | 55.9 |
| <45 μm | 1/50 | 1.96 | 2.96 | 66.2 |
| <45 μm | 1/20 | 2.09 | 3.02 | 69.2 |
| <45 μm | 1/10 | 1.99 | 3.05 | 65.3 |
| <45 μm | 1/1 | 2.14 | 3.04 | 70.4 |
| 45–75 μm | 1/100 | 2.27 | 2.95 | 77.1 |
| 45–75 μm | 1/50 | 2.17 | 2.93 | 73.4 |
| 45–75 μm | 1/20 | 2.15 | 2.96 | 72.8 |
| 45–75 μm | 1/10 | 1.89 | 3.01 | 62.3 |
| 45–75 μm | 1/1 | 1.85 | 3.10 | 59.9 |
This interpretation is also qualitatively consistent with a recent first-principles study on WB, which suggested that thermally activated ionic conduction by interlayer K+ can become feasible under moderate interlayer expansion, whereas dominant electronic band conduction is unlikely in WB. To assess within-batch specimen-to-specimen reproducibility, additional electrical-conductivity measurements were performed for representative salt-to-WB mass ratios of 1/100, 1/50, and 1/1 using independently cut specimens from the same preparation batch. As shown in Figure S2, the independently measured specimens exhibited broadly consistent temperature-dependent conductivity trends, although specimen-to-specimen variation was observed in the absolute values. Such variation may arise from local differences in density, porosity, effective current pathways, specimen geometry, and electrode contact among the cut specimens.
3.3. Evaluation of the Seebeck Coefficient
Figure shows the temperature dependence of the Seebeck coefficient (S) for WB-derived samples with different KCl–CaCl2 (3:1) addition ratios for each particle-size range. For the 45–75 μm samples (Figure a), the highest Seebeck coefficient was obtained at the lowest salt addition ratio of 1/100, reaching a maximum value of approximately 1.9 × 105 μV K–1 at 850 °C. This value is far larger than those typically observed in conventional electron-dominated thermoelectric materials and therefore requires particularly cautious interpretation. In the present study, the anomalously large thermovoltage is taken as evidence that nonelectronic contributions may be involved. However, the available data do not allow us to rigorously distinguish among ionic transport, electrode polarization, chemical-potential-driven effects, and other possible origins. Accordingly, the present manuscript does not claim definitive proof of a stable ionic thermoelectric mechanism. Rather, the XRF and XRD results indicate that K and Ca remain in the WB-derived structure after treatment, suggesting that modification of the interlayer chemical environment may be related to the observed anomalous thermovoltage.
5.
Temperature dependence of the Seebeck coefficient (S) of WB-derived materials treated with KCl–CaCl2 (3:1) at different salt addition ratios for the (a) 45–75 μm and (b) <45 μm sieve fractions. Large positive Seebeck coefficients were obtained for the 45–75 μm samples at low salt addition ratios, whereas the <45 μm samples exhibited sign reversal in certain temperature ranges.
Under the 1/100 and 1/50 salt addition conditions, the Seebeck coefficient of the 45–75 μm samples remained consistently high over the measured high-temperature range, indicating a strong correlation among particle size, salt concentration, and the layered structure in governing both the magnitude and stability of S. In contrast, the <45 μm samples (Figure b) exhibited markedly different behavior. Even under the 1/100 and 1/50 conditions, negative Seebeck coefficients were observed in certain temperature ranges, indicating sign reversal and instability of the thermoelectric response. Such polarity changes have also been reported in systems exhibiting nonelectronic thermovoltage contributions. In the present case, the sign reversal is interpreted cautiously as evidence of complex transport behavior rather than as proof of a specific ionic mechanism. The detailed origin of this response remains to be clarified by future dedicated analyses. , This behavior in the present system suggests that the measured thermovoltage may involve more than one transport contribution, although the detailed mechanism remains unresolved. The pronounced dependence on particle refinement further indicates that the Seebeck response is sensitive to processing-induced microstructural variation, but the exact origin of this response remains to be clarified.
Overall, these results indicate that particle size and molten-salt addition strongly influence the magnitude and polarity of the measured thermovoltage response. However, the present data do not establish the detailed physical origin or long-term stability of the unusually large Seebeck response, and the results should therefore be interpreted with caution.
3.4. Thermal Conductivity
Figure shows the temperature dependence of the thermal conductivity (κ) for sintered WB-derived materials prepared with different particle sizes and KCl–CaCl2 (3:1) addition ratios. In this study, κ was calculated from the measured thermal diffusivity, specific heat capacity, and bulk density. For all samples, κ exhibited either a gradual decrease or an approximately constant behavior with increasing temperature, and no anomalous enhancement was observed in the high-temperature region. Figure shows the temperature dependence of the thermal conductivity (κ) for sintered WB-derived materials prepared with different particle sizes and KCl–CaCl2 (3:1) addition ratios. In this study, κ was calculated from the measured thermal diffusivity, specific heat capacity, and bulk density. For all samples, κ exhibited either a gradual decrease or an approximately constant behavior with increasing temperature, and no anomalous enhancement was observed in the high-temperature region. This behavior indicates that heat transport in the present WB-based system is predominantly governed by lattice vibrations, with only a limited contribution from electronic carriers.
6.
Temperature dependence of the thermal conductivity (κ) of sintered WB-derived materials prepared with different KCl–CaCl2 (3:1) addition ratios for the (a) 45–75 μm and (b) <45 μm sieve fractions, measured over the temperature range of 650–850 °C.
No pronounced difference in κ was observed between the <45 μm and 45–75 μm samples. In contrast, a slight increase in κ was observed for the more densely compacted samples, which is reasonably attributed to reduced porosity and improved interparticle thermal contact. However, the magnitude of this increase remained modest. Importantly, the variation in κ was small compared with the substantial changes in electrical conductivity and Seebeck coefficient. Thus, the negative contribution of κ to the thermoelectric figure of merit ZT was relatively minor. These results indicate that thermal conductivity is not the primary performance-limiting factor in this system and that the observed ZT enhancement is mainly governed by improvements in electrical transport.
No pronounced difference in κ was observed between the <45 μm and 45–75 μm samples. In contrast, a slight increase in κ was observed for the more densely packed samples, which is reasonably attributed to reduced porosity and improved interparticle thermal contact. However, the magnitude of this increase remained modest. Importantly, the variation in κ was small compared with the substantial changes in electrical conductivity and Seebeck coefficient. Thus, the limiting effect of κ on the thermoelectric figure of merit ZT was relatively minor. These results indicate that thermal conductivity is not the primary performance-limiting factor in this system and that the observed ZT enhancement is mainly governed by improvements in charge transport and thermopower.
3.5. Thermoelectric Figure of Merit
Figure shows the temperature dependence of the thermoelectric figure of merit (ZT) for sintered WB-derived materials prepared with different particle sizes, molten-salt addition ratios, and compaction pressures. The ZT values were evaluated in the high-temperature range of 650–850 °C, and all samples exhibited an overall increase in ZT with increasing temperature. Notably, the highest value calculated using the conventional thermoelectric relation, ZT = S 2σT/κ, was 0.29 at 667 °C for the sample prepared with a particle size of 45–75 μm, a KCl–CaCl2 (3:1) addition ratio of 1/50, and a compaction pressure of 200 MPa. When evaluated using the same conventional definition, this value represents a substantial increase compared with our previously reported WB-derived system. Nevertheless, because the calculated ZT depends on the measured Seebeck coefficient, electrical conductivity, and thermal conductivity, its physical significance should be interpreted with caution in view of both the anomalously large Seebeck response and the associated experimental variability.
7.
Temperature dependence of the thermoelectric figure of merit (ZT) of sintered WB-derived materials prepared with different KCl–CaCl2 (3:1) addition ratios for the (a) 45–75 μm and (b) <45 μm sieve fractions. The 45–75 μm sample with a salt addition ratio of 1/50 exhibited the highest thermoelectric performance, reaching a maximum ZT of approximately 0.29 at 667 °C.
The increase in the conventionally calculated ZT cannot be attributed to a single measured transport parameter. Rather, it reflects the combined changes in electrical conductivity, Seebeck coefficient, and thermal conductivity induced by coordinated process optimization. Because the unusually large Seebeck response may include nonelectronic contributions, the physical significance of the calculated ZT should be interpreted cautiously. As discussed above, the intermediate particle-size range of 45–75 μm provides a favorable combination of packing behavior and measured transport response under the present processing conditions. In addition, the application of a high compaction pressure increases the measured bulk density and packing ratio. Within the scope of the present study, these observations are interpreted as evidence of coordinated process optimization rather than as direct proof of a specific microscopic transport pathway. Direct microstructural characterization, including SEM observations of fracture surfaces, grain morphology, pore structure, and interparticle interfaces, will be required in future work to clarify the microscopic origin of the observed relationship between densification behavior and electrical transport.
Importantly, the highest packing ratio does not directly produce the highest ZT. For example, the 45–75 μm sample at a salt addition ratio of 1/100 exhibited the highest packing ratio (77.1%), whereas the maximum ZT was obtained at 1/50. This result indicates that thermoelectric performance is not determined solely by densification. Instead, the results suggest that a favorable combination of transport properties is obtained through the combined influence of packing behavior, particle size, and salt-induced structural modification. In other words, a higher packing ratio or a larger salt addition does not necessarily lead to the most favorable combination of σ, S, and κ.
In contrast, samples with particle sizes below 45 μm did not exhibit comparable ZT enhancement even when relatively high packing ratios were achieved. This behavior indicates that excessive particle refinement does not provide a favorable combination of packing behavior and measured transport properties under the present processing conditions. However, the microscopic origin of this particle-size dependence remains unresolved. These results demonstrate that the maximization of ZT in the present WB-based system is governed by the cooperative optimization of particle size, compaction pressure, and molten-salt addition ratio, which collectively define the microstructural framework and transport characteristics. The conventionally calculated ZT value of 0.29 at 667 °C is therefore associated with a favorable combination of measured transport parameters under optimized processing conditions within the WB-derived system, rather than with a level of performance directly competitive with state-of-the-art high-temperature thermoelectric materials.
It should also be emphasized that the electrical conductivity of the present samples remains low, on the order of 10–5–10–4 S cm–1. Furthermore, because the anomalously large thermovoltage may include nonelectronic contributions, a conventionally calculated power factor, S 2σ, cannot necessarily be interpreted as a direct indicator of useful electrical power generation. Load-dependent output voltage, current, and power were not measured in the present study. Therefore, the practical feasibility of the present material system cannot yet be assessed from the calculated ZT alone and will require dedicated device-level measurements.
4. Conclusions
In this study, the high-temperature thermoelectric performance of natural weathered biotite (WB)-derived materials was significantly enhanced through molten-salt treatment combined with systematic optimization of processing conditions. Particular attention was paid to three key parametersthe particle size of the WB powders, the addition ratio of KCl–CaCl2 mixed salt, and the compaction pressure during pellet formationand their effects on crystal structure, densification behavior, and thermoelectric transport properties were systematically clarified.
As a result, a maximum thermoelectric figure of merit of ZT = 0.29 was achieved at 667 °C for samples prepared with a particle size of 45–75 μm, a molten-salt addition ratio of 1/50, and a compaction pressure of 200 MPa. This value represents a substantial improvement over the previously reported ZT of 0.015 at 747 °C for the same material system. The performance enhancement is associated with the cooperative optimization of particle size, salt addition, and compaction pressure, which collectively improve densification behavior and modify the measured transport response of the sintered body. The present study therefore provides process–structure–property insight, although direct mechanistic determination will require additional microstructural and transport-specific analyses. The results further suggest that modification of the interlayer chemical/structural environment may contribute to the anomalous thermovoltage response of this system; however, the detailed origin of this response has not yet been conclusively established.
These findings demonstrate that the measured transport response in WB-derived materials is governed by a well-defined process–structure–property relationship based on the interplay among particle size, densification, and transport behavior. In the present study, the reported ZT values are derived from the conventional thermoelectric formalism and should be interpreted with caution in view of the anomalously large Seebeck response. The XRD results indicate that the WB-derived layered framework is largely retained after molten-salt treatment, while additional diffraction features suggest reaction-induced structural modifications; however, exact phase identification remains unresolved. The present study provides process–structure–property insight into a natural mineral-derived layered silicate system, although further work is needed to determine its practical feasibility for device applications. This study therefore provides useful design insights into high-temperature transport response in natural layered silicate-based materials in the range of 600–900 °C. Future work will focus on long-term stability under continuous thermal gradients, rigorous clarification of the anomalous thermovoltage response, and direct evaluation of load-dependent voltage, current, and output power to assess the practical applicability of this material system.
Supplementary Material
Acknowledgments
This work was carried out at the Japan Atomic Energy Agency (JAEA) and was financially supported by the JAEA Fund for Exploratory Researches (Houga fund). The authors also acknowledge financial support from JSPS KAKENHI (Grants-in-Aid for Scientific Research), Grant Number 22K19001.
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c05333.
(Figure S1) XRF spectra of WB-derived materials prepared with different particle-size fractions and KCl–CaCl2 addition ratios. (Figure S2) Additional electrical-conductivity measurements for representative salt-to-WB mass ratios of 1/100, 1/50, and 1/1 were added to evaluate within-batch specimen-to-specimen reproducibility (PDF)
The authors declare no competing financial interest.
References
- Papapetrou M., Kosmadakis G., Cipollina A., La Commare U., Micale G.. Industrial Waste Heat: Estimation of the Technically Available Resource in the EU per Industrial Sector, Temperature Level and Country. Appl. Therm. Eng. 2018;138(February):207–216. doi: 10.1016/j.applthermaleng.2018.04.043. [DOI] [Google Scholar]
- Bell L. E.. Cooling, Heating, Generating Power, and Recovering Waste Heat with Thermoelectric Systems. Science. 2008;321(5895):1457–1461. doi: 10.1126/science.1158899. [DOI] [PubMed] [Google Scholar]
- Snyder G. J., Toberer E. S.. Complex TE Materials. Nat. Mater. 2008;7(February):105–114. doi: 10.1038/nmat2090. [DOI] [PubMed] [Google Scholar]
- Wood C.. Materials for Thermoelectric Energy Conversion. Rep. Prog. Phys. 1988;51:459–539. doi: 10.1088/0034-4885/51/4/001. [DOI] [Google Scholar]
- Rowe, D. M. CRC Handbook of Thermoelectrics; CRC Press, 1995. [Google Scholar]
- Honda M., Okamoto Y., Shimoyama I., Shiwaku H., Suzuki S., Yaita T.. Mechanism of Cs Removal from Fukushima Weathered Biotite by Heat Treatment with a NaCl–CaCl 2 Mixed Salt. ACS Omega. 2017;2(2):721–727. doi: 10.1021/acsomega.6b00372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Honda M., Shimoyama I., Kogure T., Baba Y., Suzuki S., Yaita T.. Proposed Cesium-Free Mineralization Method for Soil Decontamination: Demonstration of Cesium Removal from Weathered Biotite. ACS Omega. 2017;2(12):8678–8681. doi: 10.1021/acsomega.7b01304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Honda M., Kaneta Y., Muraguchi M., Hayakawa K., Oda M., Iino C., Ishii H., Goto T.. Sustainable Thermoelectric Materials: Utilizing Fukushima Weathered Biotite via Molten Salt Treatment. AIP Adv. 2024;14:055034. doi: 10.1063/5.0206317. [DOI] [Google Scholar]
- Nandihalli N., Liang R., Wijethunge D., Zhou N., Kleinke H., Nandihalli N., Liang R., Wijethunge D.. Composites with NiSb Nanocoating. AIP Adv. 2018;125304:0–12. doi: 10.1063/1.5038675. [DOI] [Google Scholar]
- Sulaiman S., Izman S., Uday M. B., Omar M. F.. Review on Grain Size e Ff Ects on Thermal Conductivity in ZnO Thermoelectric Materials. RSC Adv. 2022;12:5428–5438. doi: 10.1039/D1RA06133J. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liang W., Rabin O., Hochbaum A. I., Fardy M., Zhang M., Yang P.. Thermoelectric Properties of P-Type PbSe Nanowires. Nano Res. 2009;2:394–399. doi: 10.1007/s12274-009-9039-2. [DOI] [Google Scholar]
- Roh J. W., Hippalgaonkar K., Ham J. H., Chen R., Li M. Z., Ercius P., Majumdar A., Kim W., Lee W.. Observation of Anisotropy in Thermal Conductivity of Individual Single- Crystalline Bismuth Nanowires. ACS Nano. 2011;5(5):3954–3960. doi: 10.1021/nn200474d. [DOI] [PubMed] [Google Scholar]
- Ioannou I., Ioannou P. S., Kyratsi T., Giapintzakis J.. Effect of Starting Powder Particle Size on the Thermoelectric Properties of Hot-Pressed Bi0.3Sb1.7Te3 Alloys. Materials. 2024;17:318. doi: 10.3390/ma17020318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schrade M., Berland K., Eliassen S. N. H., Guzik M. N., Echevarria-Bonet C., Sørby M. H., Jenuš P., Hauback B. C., Tofan R., Gunnæs A. E.. et al. The Role of Grain Boundary Scattering in Reducing the Thermal Conductivity of Polycrystalline X NiSn (X = Hf, Zr, Ti) Half-Heusler Alloys. Sci. Rep. 2017;7:13760. doi: 10.1038/s41598-017-14013-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Feng B., Li G., Pan Z., Xiaoming H., Peihai L., Zhu H., Yawei L., Fan X.. Effect of Synthesis Processes on the Thermoelectric Properties of BiCuSeO Oxyselenides. J. Alloys Compd. 2018;754:131–138. doi: 10.1016/j.jallcom.2018.04.307. [DOI] [Google Scholar]
- Chiu W., Chen C., Chen Y.. A Strategy to Optimize the Thermoelectric Performance in a Spark Plasma Sintering Process. Sci. Rep. 2016;6:1–9. doi: 10.1038/srep23143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yabuta R., Oda M., Muraguchi M., Ishii H., Hayakawa K., Honda M.. Elucidation of the ionic conductivity mechanism of weathered biotite using first-principles calculation. Jpn. J. Appl. Phys. 2026;65:024001. doi: 10.35848/1347-4065/ae2c2d. [DOI] [Google Scholar]
- Zhao D., Martinelli A., Willfahrt A., Fischer T., Bernin D., Khan Z. U., Shahi M., Brill J., Jonsson M. P., Fabiano S., Crispin X.. Polymer Gels with Tunable Ionic Seebeck Coefficient for Ultra-Sensitive Printed Thermopiles. Nat. Commun. 2019;10:1–8. doi: 10.1038/s41467-019-08930-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Goracci G., Ogundiran M. B., Barzegar M., Iturrospe A., Arbe A., Dolado J. S.. Kaolin Clay-Based Geopolymer for Ionic Thermoelectric Energy Harvesting. ACS Omega. 2024;9(12):13728–13737. doi: 10.1021/acsomega.3c08257. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.