Solid‐State Electrochemical Thermal Transistors

Thermal transistors that electrically control heat flow have attracted growing attention as thermal management devices and phonon logic circuits. Although several thermal transistors are demonstrated, the use of liquid electrolytes may limit the application from the viewpoint of reliability or liquid leakage. Herein, a solid‐state thermal transistor that can electrochemically control the heat flow with an on‐to‐off ratio of the thermal conductivity (κ) of ≈4 without using any liquid is demonstrated. The thermal transistor is a multilayer film composed of an upper electrode, strontium cobaltite (SrCoOx), solid electrolyte, and bottom electrode. An electrochemical redox treatment at 280 °C in air repeatedly modulates the crystal structure and κ of the SrCoOx layer. The fully oxidized perovskite‐structured SrCoO3 layer shows a high κ ≈3 .8 W m−1 K−1, whereas the fully reduced defect perovskite‐structured SrCoO2 layer shows a low κ ≈ 0.95 W m−1 K−1. The present solid‐state electrochemical thermal transistor may become next‐generation devices toward future thermal management technology.

The active material shows nonlinear heat transport properties upon a phase change, while the switching material controls the phase of the active material repeatedly. Thus, switching electrically modulates the thermal conductivity (κ) of the active material.
History of thermal transistor development is short, only less than 10 years. In 2014, Ben-Abdallah and Biehs theoretically predicted a thermal transistor that uses the metal-to-insulator transition (MIT) of VO 2 [7] as the active layer. According to the Wiedemann-Frantz law, metallic VO 2 exhibits κ that is at least ≈1 W m −1 K −1 higher than the insulator VO 2 due to the electron contribution. That is, κ ele = L·σ·T, where L is the Lorentz number (2.44 × 10 −8 W Ω K −2 ), σ is the electrical conductivity (≈1500 S cm −1 for metal phase VO 2 [8] ), and T is the absolute temperature. Since the MIT occurs at ≈68 °C in the bulk state, they predicted that κ of the VO 2 layer can be switched by Joule heating of the VO 2 layer above 68 °C. However, in 2017, Lee et al. [9] reported that the Wiedemann-Franz law was not applicable to VO 2 and the κ values of VO 2 before and after the MIT were almost the same [10] . Although Zhu et al. [11] demonstrated a thermal rectification in VO 2 beams, the value is not sufficient for thermal transistor applications. Consequently, the thermal transistor described in ref. [7] is impossible to realize.
Recently, electrochemical control of κ for materials using liquid electrolytes has attracted attention. Cho et al. [12] realized the first thermal transistor in 2014. They measured κ of a LiCoO 2 thin film before and after electrochemical delithiation using the time-domain thermo-reflectance (TDTR) method. κ of the LiCoO 2 film can be reversibly modulated in the range of ≈5.4-3.7 W m −1 K −1 . To date, several similar thermal transistors have been proposed. In 2018, Sood et al. [13] demonstrated the electrochemical Li intercalation/deintercalation using a liquid electrolyte modulates κ of MoS 2 . In 2020, transition metal oxide (TMO)-based thermal transistors using an ionic liquid as the electrolyte have been reported. [14,15] Lu et al. [15] used SrCoO x as the active layer. The κ ratio of oxidized SrCoO 3 (κ = 4.33 ±1.62 W m −1 K −1 ) versus protonated HSrCoO 2.5 (κ = 0.44 ± 0.06 W m −1 K −1 ) is 10 ± 4. Very recently, Zhou et al. [16] reported a heat conductor-insulator transition in electrochemical hybrid superlattices composed of MoS 2 and an organic molecule. These thermal transistors utilize the change in κ of the active material when ions are intercalated/deintercalated electrochemically. Although these materials show suitable thermal transistor characteristics, the use of liquids (electrolyte,

Introduction
Thermal transistors, [1][2][3] which can electrically control heat flow, have attracted growing attention as thermal management devices [4] and phonon logic circuits. [5,6] A thermal transistor is composed of an active material and a switching material.
ionic liquid) may limit the application because such devices must be placed in containers and sealed. Thus, the development of solid-state thermal transistors is crucial.
To realize solid-state thermal transistors, we used a solid electrolyte, yttria-stabilized zirconia (YSZ), though the operation temperature is a bit high (≈300 °C) compared to the liquid electrolyte. We choose YSZ from the following reasons. YSZ is an oxide (O 2− ) ion conductor and has been applied as the solid electrolyte for solid oxide fuel cells. [17] Additionally, rather large-sized YSZ single crystals are commercially available. Most importantly, many perovskite-related TMOs are heteroepitaxially grown on YSZ single-crystal substrates. [18][19][20] In this study, we focus on SrCoO x (2 ≤ x ≤ 3) as the active material for a solid-state electrochemical thermal transistor. SrCoO x is known as an oxygen sponge, and the O 2− ion concentration of SrCoO x can be controlled at relatively low temperatures. [21][22][23][24] Our preliminary studies (Supporting Information Section S1, Figures S1−S6, Table S1) reveal that a SrCoO 2.5 film with a brownmillerite (BM) structure grown on YSZ is electrochemically oxidized into a perovskite (P-) SrCoO 3 film, which shows a high σ of ≈1400 S cm −1 . [25] Moreover, we found that a SrCoO 2.5 film on YSZ can be electrochemically reduced into a defect perovskite (DP-) SrCoO 2 film that is highly insulating. Generally, κ of material is expressed as the summation of κ due to the quantized lattice vibration (phonon) and κ due to the electron. [26] The former depends on the crystal structure and quality. The latter is given by the Wiedemann-Franz law as described above. From these observations, we expect that κ of P-SrCoO 3 is higher than that of DP-SrCoO 2 , as schematically illustrated in Figure 1.
Here, we demonstrate a solid-state electrochemical thermal transistor. The thermal transistor consists of a multilayered structure composed of a Pt upper electrode, a SrCoO 2.5 active layer, a solid electrolyte Gd-doped CeO 2 layer on a YSZ substrate, and a Pt bottom electrode on the backside of the YSZ. The multilayer sample is placed on a heater stage and heated to 280 °C in air. Then electrochemical oxidation/ reduction treatment at 280 °C in air repeatedly modulates the crystal structure and κ of the SrCoO x layer. The fully oxidized P-SrCoO 3 layer shows a high κ ≈3.8 W m −1 K −1 . By contrast, the fully reduced DP-SrCoO 2 layer shows a low κ ≈0.95 W m −1 K −1 . Consequently, a solid-state thermal transistor electrochemically controls the heat flow with an on-to-off κ ratio of ≈4.

Results and Discussion
First, we fabricated solid-state electrochemical thermal transistors (Supporting Information Section S2, Figures S7 and S8). Then, a fabricated solid-state electrochemical thermal transistor (5 mm × 5 mm) was set on a Pt-coated glass substrate (Supporting Information Section S3, Figure S9). After heating to 280 °C in air, an electrochemical redox treatment was performed by applying a constant current of −50 µA for reduction and +50 µA for oxidation. We controlled the current application time by monitoring the flown electron density Q = (I·t)/(e·V), where I is the flown current, t is the applied time, e is the electron charge, and V is the volume of the SrCoO 2.5 film (5 mm × 5 mm × ≈50 nm). After applying the current, the device was immediately cooled to room temperature.
The electrochemical redox treatment began by applying a negative current to reduce BM-SrCoO 2.5 into DP-SrCoO 2 (Figure 2a). Initially, the DC voltage in the electrochemical reduction was ≈−5.1 V, which was close to that expected from the DC resistance of the YSZ substrate (≈100 kΩ, 5 mm × 5 mm × 0.5 mm) (Supporting Information Section S3, Figure S10). Two semicircles of ≈100 and ≈25 kΩ were observed when we measured the impedance spectroscopy (Cole-Cole plot, data not shown). The former indicates resistance of YSZ substrate and the latter indicates the YSZ/ Pt interface resistance. As Q increased, the absolute value of the voltage gradually increased and became saturated around −5.7 V. Then it jumped to −6.1 V and became saturated again when Q was ≈1.7 × 10 22     Electrochemical reduction of GDC [27] occurred after the overreduction treatment (data not shown). It should be noted that GDC is easily oxidized during electrochemical oxidation.
Separately, we measured the κ of the SrCoO x layer by the TDTR method using the top Pt electrode as the transducer (Figures S15 and S16, Supporting Information). The TDTR measurement was performed at room temperature in air. Figure 2b shows the TDTR decay curves upon electrochemical reduction (see also Figure S17, Supporting Information). Compared to the as-grown state (0 cm −3 ), the decay slowed as Q increased, suggesting a decrease of κ of the SrCoO x layer. Figure 2c plots the estimated κ as a function of Q. κ of as-grown BM-SrCoO 2.5 was ≈1.9 W m −1 K −1 but gradually decreased with Q. When Q reached ≈1.7 × 10 22 cm −3 , κ became saturated at ≈1 W m −1 K −1 .
Next the device was oxidized electrochemically (Figure 2d). The initial DC voltage was ≈4.0 V. As Q increased, the voltage increased gradually and became saturated at ≈4.2 V. When Q exceeded ≈1.7 × 10 22 cm −3 , the voltage dramatically increased and became saturated at ≈4.8 V when Q reached ≈3.4 × 10 22 cm −3 .
The change in the crystalline phases is as follows (2) The XRD patterns confirmed these changes (Figure 3c-e). These results reveal that the x value in SrCoO x can be controlled electrochemically between 2 and 3 without destroying the crystal structure.
The TDTR decay became faster as Q increased (Figure 2e), suggesting κ of the SrCoO x layer increased. Figure 2f shows two stepwise increases of κ of the SrCoO x at Q ≈1.7 × 10 22 and ≈3.4 × 10 22 cm −3 . These stepwise increases of κ correspond to the oxidation from DP-SrCoO 2 to BM-SrCoO 2.5 and from BM-SrCoO 2.5 to P-SrCoO 3 . The κ of fully oxidized P-SrCoO 3 was ≈3.5 W m −1 K −1 . These results reveal that κ of the SrCoO x layer can be controlled electrochemically between ≈1 and ≈3.5 W m −1 K −1 using YSZ as the solid electrolyte.
Finally, we examined the repeatability of the thermal transistor. Figure 4 shows the change in the crystal lattice of SrCoO x layer on the redox treatment. In the out-of-plane XRD patterns, the diffraction peak around 5.05 nm −1 was 008 for the as-grown BM-SrCoO 2.5 . The reduction treatment shifted the diffraction peak to ≈5.4 nm −1 (002 DP-SrCoO 2 ), while the oxidation treatment shifted it to ≈5.25 nm −1 (002 P-SrCoO 3 ). Repeating the redox treatments ten times did not change the peak position or shape except for the first reduction and oxidation treatments. Similarly, the applied voltage was almost unchanged in each redox cycle except for the first reduction and oxidation treatments ( Figure S11, Supporting Information). Figure 4b shows the change in the d-value of reduced DP-SrCoO 2 and oxidized P-SrCoO 3 layers upon redox cycles. The extracted lattice parameter c was 0.3706 nm for reduced DP-SrCoO 2 and 0.3795 nm for oxidized P-SrCoO 3 . These results reveal that the crystal structure is maintained after redox cycling. Figure 5 shows the change in the TDTR decay curves of the redox-cycled device. The TDTR decay of oxidized P-SrCoO 3 was always faster than that of reduced DP-SrCoO 2 . Using these TDTR decay curves, the change in κ of reduced DP-SrCoO 2 and oxidized P-SrCoO 3 layers was simulated in the redox cycles (see Figure S18, Supporting Information). The average κ values were 0.95 W m −1 K −1 for reduced DP-SrCoO 2 and 3.8 W m −1 K −1 for oxidized P-SrCoO 3 , indicating an on-tooff thermal conductivity ratio for the SrCoO x layer of 4. If we assume the electron contribution of the κ ele of P-SrCoO 3 in the cross-plane direction by the Wiedemann-Frantz law using the in-plane electrical conductivity (≈590 S cm −1 , Section S1.3, Supporting Information), the value is ≈0.43 W m −1 K −1 (≈11% of the observed κ). These results indicate the on-to-off thermal conductivity ratio mainly comes from the difference between DP and P structure of SrCoO x .
Adv. Funct. Mater. 2023, 33,   Here, we compare the characteristics of the present solidstate electrochemical thermal transistor with the reported liquid-based electrochemical thermal transistors ( Table 1). The present solid-state thermal transistor shows comparable characteristics except that the operating temperature is high due to low oxide ion conductivity of YSZ crystal. In order to overcome this problem, reduction of the resistance of the solid electrolyte is crucial.
If we except the problem of high operating temperature, the present solid-state electrochemical thermal transistor has several advantages compared to liquid-based electrochemical thermal transistors. First, it does not need to be placed in a container and sealed. Second, it shows stable operations after repeated cycles. Third, it has a good reproducibility. The multilayered structure composed of Pt, SrCoO x , GDC, and YSZ is maintained after the redox cycles (Figures S12 and S13, Supporting Information), and the crystal structure of SrCoO x maintains the perovskite structure upon the redox reaction ( Figure S14, Supporting Information). We fabricated many thermal transistors (>20 pieces) and tested the cycle properties. And we confirmed that all the devices show the on/off ratio of ≈4 between the fully oxidized perovskite SrCoO 3 versus the reduced defect perovskite SrCoO 2 . Fourth, the device operations obey Faraday's law of electrolysis without current leakage. Thus, controlling Q can realize on/off control of the device. In the present device, the solid electrolyte (0.5 mm thick YSZ crystal) and the solid electrolyte/Pt interface dominate the current flow. In other word, use of a solid electrolyte with a higher O 2− ion conductivity or reducing the thickness of a solid electrolyte is effective to reduce the operating temperature and switching time of the device.

Conclusions
In summary, we demonstrated a solid-state electrochemical thermal transistor without any liquid that can control the heat flow using the change in the thermal conductivity of the SrCoO x layer. The on-to-off ratio of the thermal conductivity (κ) was ≈4. The solid-state electrochemical thermal transistor was composed of the upper electrode, SrCoO x , the solid electrolyte, and the bottom electrode. The electrochemical redox treatment at 280 °C in air turned the thermal transistor on and off and repeatedly modulates the crystal structure and κ of the SrCoO x layer. When the thermal transistor was on, κ of the SrCoO 3 layer was ≈3.8 W m −1 K −1 . By contrast, κ of the SrCoO 2 layer was ≈0.95 W m −1 K −1 when it was off. Additionally, we confirmed the cycle properties of the thermal transistor (10 cycles).
Although the present solid-state thermal transistor shows comparable characteristics with liquid-based thermal transistors ever reported, the operating temperature is high due to low oxide ion conductivity of YSZ crystal. In order to develop