This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Present Address: ]NTT Secure Platform Laboratories, NTT Corporation, Musashino 180-8585, Japan

Present Address: ]Integrated Research for Energy and Environment Advanced Technology, Kyushu Institute of Technology, Kitakyushu, Fukuoka 804-8550, Japan

Phenomenological analysis of transverse thermoelectric generation and cooling performance in magnetic/thermoelectric hybrid systems

Kaoru Yamamoto [ [email protected] Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan    Ryo Iguchi Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan    Asuka Miura [ Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan    Weinan Zhou Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan    Yuya Sakuraba Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan PRESTO, Japan Science and Technology Agency, Saitama 332-0012, Japan    Yoshio Miura Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan Center for Spintronics Research Network, Osaka University, Osaka 560-8531, Japan    Ken-ichi Uchida [email protected] Research Center for Magnetic and Spintronic Materials, National Institute for Materials Science (NIMS), Tsukuba 305-0047, Japan Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan Center for Spintronics Research Network, Tohoku University, Sendai 980-8577, Japan
Abstract

We phenomenologically calculate the performance of the recently-observed Seebeck-driven transverse thermoelectric generation (STTG) for various systems in terms of the thermopower, power factor, and figure of merit to demonstrate the usefulness of STTG. The STTG system consists of a closed circuit comprising thermoelectric and magnetic materials which exhibit the Seebeck and anomalous Hall effects, respectively. When a temperature gradient is applied to the hybrid system, the Seebeck effect in the thermoelectric material layer generates a longitudinal charge current in the closed circuit and the charge current subsequently drives the anomalous Hall effect in the magnetic material layer. The anomalous Hall voltage driven by the Seebeck effect has a similar symmetry to the transverse thermoelectric conversion based on the anomalous Nernst effect. We find that the thermoelectric properties of STTG can be much better than those of the anomalous Nernst effect by increasing the Seebeck coefficient and anomalous Hall angle of the thermoelectric and magnetic materials, respectively, as well as by optimizing their dimensions. We also formulate the electronic cooling performance in the STTG system, confirming the reciprocal relation for the hybrid transverse thermoelectric conversion.

I Introduction

Thermoelectric power generation based on transverse thermoelectric phenomena has been actively investigated in spin caloritronics.Bauer2012NatMaterReview ; Boona2014Ene&EngSciReview ; Uchida2016ProcIEEESSEReview ; Mizuguchi2019STAMreview ; Uchida2021APL In particular, the anomalous Nernst effect (ANE), in which the thermoelectric voltage appears perpendicular to the direction of a temperature gradient and magnetization in magnetic materials, has gained much interest owing to its physical mechanism and thermoelectric functionalities. ANE enables thermoelectric generation with a simple thermopile structure and a convenient scaling behavior.Sakuraba2013APExThermopile ; Sakuraba2015ScrMaterANETEG ; Yang2017AIPAdvCoiledWire By utilizing the features of ANE, a coil-shaped thermoelectric generator and a flexible heat flux sensor have recently been demonstrated.Yang2017AIPAdvCoiledWire ; ZhouAPL2020heatflux

For applications of the transverse thermoelectric generation, a large transverse thermopower of >20μK1>20\ \mu\text{V }\text{K}^{-1} is at least necessary according to the estimation in Ref. Sakuraba2015ScrMaterANETEG, . To realize such large transverse thermopower, ANE has been investigated in various materials including ferromagnetic alloys,Mizuguchi2012APEXFePtFilm ; Hasegawa2015APLOrderedAlloyFilm ; Isogami2017APExFe4Nfilm ; Seki2018JphysDFePtfilm ; Nakayama2019PRMateFeGa ; Sakai2020NatureFe3Ga Heusler compounds,Sakai2018NatPhysCo2MnGa ; Reichlova2018APLCo2MnGafilm ; Park2020PRBCMGthikcness ; Sakuraba2020PRBCMAS ; SakurabaCMG and permanent magnets.Miura2019APLSmCo5 ; MiuraAPL2020Tempdep Multilayer films may produce large transverse thermoelectric voltage with increasing the number of layers.Uchida2015PRBMultiLayer ; Fang2016PRBPtComultilayer ; Ramos2019APLFe3O4/Pt ; Seki2021PRB However, the obtained transverse thermopower is still much smaller than 10μK110\ \mu\text{V }\text{K}^{-1}; further materials exploration and device engineering are necessary to obtain larger transverse thermopower.

Recently, Zhou et al. proposed and experimentally demonstrated a transverse thermoelectric generation with a similar symmetry to ANE but a different driving principle from ANE. Zou2020 The device used in this experiment consists of a closed circuit comprising thermoelectric and magnetic materials which show the Seebeck effect and anomalous Hall effect (AHE), respectively. When a temperature gradient is applied to the hybrid structure in the xx direction, the Seebeck effect in the thermoelectric material layer generates a charge current in the closed circuit and the charge current subsequently drives AHE in the magnetic material layer [see Fig. 1(a)]. The anomalous Hall voltage driven by the Seebeck-effect-induced charge current has the same symmetry as the transverse thermoelectric conversion based on ANE when the magnetization 𝐌\mathbf{M} is along the zz direction, boosting the transverse thermopower in the magnetic layer. Zhou et al. observed a giant transverse thermopower of 82.3 μK1\mu\text{V }\text{K}^{-1} in a Co2MnGa\text{Co}_{2}\text{MnGa}/n-type Si hybrid structure, which is more than 10 times larger than the anomalous Nernst coefficient of a Co2MnGa\text{Co}_{2}\text{MnGa} monolayer. The observed effect is referred to as the Seebeck-driven transverse thermoelectric generation (STTG).Zou2020 They also demonstrated that the sign of the transverse thermopower induced by STTG can be changed by reversing the sign of the Seebeck coefficient of the thermoelectric material layer; a negative transverse thermopower of 41.0-41.0 μK1\mu\text{V }\text{K}^{-1} was observed in a Co2MnGa\text{Co}_{2}\text{MnGa}/p-type Si structure.Zou2020 The previous work, however, focused only on a few combinations of thermoelectric and magnetic materials.

In this study, we phenomenologically calculate the thermopower, power factor, and figure of merit for STTG for various combinations of thermoelectric and magnetic materials to demonstrate the usefulness of STTG. The STTG system can exhibit much better performance than existing ANE materials by optimizing transport properties and dimensions of the thermoelectric and magnetic materials. Our results thus give strategies to realize efficient STTG-based energy harvesting and heat sensing devices. To confirm the reciprocal relation for the hybrid transverse thermoelectric conversion, we also formulate the electronic cooling performance in the STTG system.

Refer to caption
Figure 1: (a) A schematic illustration of the STTG system. (b) Equivalent circuits of the STTG system in the xx (left) and yy (right) directions. Battery symbols in the left circuit denote the electromotive force generated by the Seebeck effects in the thermoelectric and magnetic materials, while the symbol in the right circuit denotes the electromotive force generated by the ANE and Seebeck-driven AHE. The internal resistance in the magnetic material in the xx direction includes the feedback effect due to AHE.
Refer to caption
Figure 2: The size ratio rr dependence of the transverse thermopower StotyS_{\text{tot}}^{y} [Eq. (5)] for various ferromagnetic metals (FMs) [(a) Ni, (b) Co2MnGa\text{Co}_{2}\text{MnGa}, (c) FePt, (d) NdFeB, and (e) SmCo5\text{SmCo}_{5}] connected to typical thermoelectric materials (n-type Si(1), n-type Si(2), p-type Si, and Bi2Te3\text{Bi}_{2}\text{Te}_{3}). The inset to (b) shows the tanθH\tan\theta_{\text{H}} dependence of StotyS_{\text{tot}}^{y} for the Co2MnGa\text{Co}_{2}\text{MnGa}/n-Si(1) junction, where tanθH\tan\theta_{\text{H}} is treated as a parameter with fixing ρM\rho_{\text{M}} at r=105r=10^{5}.

II Model and setup

The following phenomenological equations for the electric field 𝐄\mathbf{E} and heat current density 𝐪\mathbf{q} describes the transport properties in the STTG system:Harman1962JAPTheoryNernst ; Landau ; Uchida2016ProcIEEESSEReview

𝐄\displaystyle\mathbf{E} =ρ𝐣+ST+ρAHE(𝐦×𝐣)SANE(𝐦×T),\displaystyle=\rho\mathbf{j}+S\nabla T+\rho_{\text{AHE}}(\mathbf{m}\times\mathbf{j})-S_{\text{ANE}}(\mathbf{m}\times\nabla T), (1)
𝐪\displaystyle\mathbf{q} =ST𝐣κTSANET(𝐦×𝐣)+κRL(𝐦×T),\displaystyle=ST\mathbf{j}-\kappa\nabla T-S_{\text{ANE}}T(\mathbf{m}\times\mathbf{j})+\kappa_{\text{RL}}(\mathbf{m}\times\nabla T), (2)

where 𝐣\mathbf{j}, T\nabla T, and 𝐦\mathbf{m} are the charge current density, the temperature gradient, and the unit vector along 𝐌\mathbf{M}, respectively. ρ=ρxx\rho=\rho_{xx} is the longitudinal resistivity, SS the Seebeck coefficient, κ\kappa the thermal conductivity, ρAHE=ρyx\rho_{\text{AHE}}=\rho_{yx} the anomalous Hall resistivity, SANES_{\text{ANE}} the anomalous Nernst coefficient, and κRL\kappa_{\text{RL}} the Righi-Leduc coefficient.Zhang2000LeducRighi1 ; Li2017LeducRighi2 Here, we neglect the thickness dependence of the transport coefficients and the magnetic field and magnetization dependences of ρ\rho, SS, and κ\kappa for simplicity. Note that this formalism can also be applied to the ordinary transverse transport phenomena by replacing 𝐦\mathbf{m} in Eqs. (1) and (2) with a magnetic field. In our model calculation, we assume that the STTG system is in the isothermal condition in the yy direction, yT=0\nabla_{y}T=0,Harman1962JAPTheoryNernst ; Uchida2016ProcIEEESSEReview and that charge and heat currents in the zz direction are perfectly blocked by an ideal insulator. Thus, the following calculation is independent of the dimensions of the insulator layer depicted in Fig. 1(a). In practice, the insulator layer should be as thin as possible because a heat current in the layer does not contribute to the thermoelectric conversion. In the following calculations, for simplicity, we neglect the interface effect at the junctions of each layer, although the interface effect may affect the transport properties when the thicknesses of the thermoelectric and magnetic layers are small.Uchida2015PRBMultiLayer ; Fang2016PRBPtComultilayer ; Ramos2019APLFe3O4/Pt ; Seki2021PRB

To determine the temperature gradients in the thermoelectric and magnetic materials, we solve 𝐪=𝐄𝐣\nabla\cdot\mathbf{q}=\mathbf{E}\cdot\mathbf{j} with the following boundary conditions: TTE(M)=ThT_{\text{TE(M)}}=T_{\text{h}} at x=0x=0 and TTE(M)=TcT_{\text{TE(M)}}=T_{\text{c}} at x=LTE(M)xx=L_{\text{TE(M)}}^{x}. Here, TTE(M)T_{\text{TE(M)}} is the temperature in the thermoelectric (magnetic) material, Th(c)T_{\text{h(c)}} is the temperature of the hot (cold) reservoir, and LTE(M)xL_{\text{TE(M)}}^{x} is the length of the thermoelectric (magnetic) material in the xx direction. The notations for the dimensions along the yy and zz directions are defined in a similar manner. We set LTEx=LMx=LxL_{\text{TE}}^{x}=L_{\text{M}}^{x}=L^{x} following the configuration depicted in Fig. 1(a). With the above conditions, we obtain

xTTE\displaystyle\nabla_{x}T_{\text{TE}} =ΔTLx+ρTE(jTEx)22κTE(Lx2x),\displaystyle=-\frac{\Delta T}{L^{x}}+\frac{\rho_{\text{TE}}(j_{\text{TE}}^{x})^{2}}{2\kappa_{\text{TE}}}(L^{x}-2x), (3)
xTM\displaystyle\nabla_{x}T_{\text{M}} =ΔTLx+{ρM[(jMx)2+(jMy)2]2κM\displaystyle=-\frac{\Delta T}{L^{x}}+\left\{\frac{\rho_{\text{M}}[(j_{\text{M}}^{x})^{2}+(j_{\text{M}}^{y})^{2}]}{2\kappa_{\text{M}}}\right.
+ΔTLxSANEjMyκM}(Lx2x),\displaystyle\left.+\frac{\Delta T}{L^{x}}\frac{S_{\text{ANE}}j_{\text{M}}^{y}}{\kappa_{\text{M}}}\right\}(L^{x}-2x), (4)

where we assumed x(y)jMx(y)=0\nabla_{x(y)}j_{\text{M}}^{x(y)}=0 and SANEjMyLMx/κM1S_{\text{ANE}}j_{\text{M}}^{y}L_{\text{M}}^{x}/\kappa_{\text{M}}\ll 1. Here, ρTE(M)\rho_{\text{TE(M)}}, κTE(M)\kappa_{\text{TE(M)}}, and jTE(M)xj_{\text{TE(M)}}^{x} are the resistivity, the thermal conductivity, and the charge current density of the thermoelectric (magnetic) material, respectively, jMyj_{\text{M}}^{y} is the charge current density in the magnetic material in the yy direction, and ΔT=ThTc\Delta T=T_{\text{h}}-T_{\text{c}}.

To calculate the thermoelectric performance in the STTG system, we need to solve Eq. (1) with the boundary conditions obtained from the equivalent circuits shown in Fig. 1(b): ITEx+IMx=0I_{\text{TE}}^{x}+I_{\text{M}}^{x}=0, VTEx=VMxV_{\text{TE}}^{x}=V_{\text{M}}^{x}, and VMy=RloadIMyV_{\text{M}}^{y}=R_{\text{load}}I_{\text{M}}^{y}, where ITE(M)xI_{\text{TE}(\text{M})}^{x} and ITE(M)yI_{\text{TE}(\text{M})}^{y} are the charge currents in the thermoelectric (magnetic) material in the xx and yy directions, respectively, VTE(M)x0LTE(M)xETE(M)x𝑑xV_{\text{TE}(\text{M})}^{x}\equiv-\int_{0}^{L_{\text{TE}(\text{M})}^{x}}E_{\text{TE}(\text{M})}^{x}dx is the voltage in the thermoelectric (magnetic) material in the xx direction, VMy0LMyEMy𝑑yV_{\text{M}}^{y}\equiv-\int_{0}^{L_{\text{M}}^{y}}E_{\text{M}}^{y}dy is the voltage in the magnetic material in the yy direction, ETE(M)xE_{\text{TE}(\text{M})}^{x} and EMyE_{\text{M}}^{y} are the electric fields, and RloadR_{\text{load}} is the load resistance. In the following calculation, we use the value of VMyV_{\text{M}}^{y} at x=LMx/2x=L_{\text{M}}^{x}/2.Harman1962JAPTheoryNernst

III Results and discussions

III.1 Transverse thermopower

By solving Eq. (1) with the aforementioned boundary conditions, we obtain the transverse thermopower for the STTG system as

Stoty(EMyxT)Rload=SANE+ρAHEρTE/r+ρM(SMSTE),\displaystyle S_{\text{tot}}^{y}\equiv\left(\frac{E^{y}_{\text{M}}}{-\nabla_{x}T}\right)_{R_{\text{load}}\to\infty}=S_{\text{ANE}}+\frac{\rho_{\text{AHE}}}{\rho_{\text{TE}}/r+\rho_{\text{M}}}(S_{\text{M}}-S_{\text{TE}}), (5)

where STE(M)S_{\text{TE(M)}} is the Seebeck coefficient of the thermoelectric (magnetic) material and rLTEyLTEz/(LMyLMz)r\equiv L_{\text{TE}}^{y}L_{\text{TE}}^{z}/(L_{\text{M}}^{y}L_{\text{M}}^{z}) is the size ratio between the thermoelectric and magnetic materials; see Methods in Ref. Zou2020, for further details of the derivation. Equation (5) shows that the transverse thermopower for the STTG system can be enhanced owing to the superposition of the ANE contribution in the magnetic layer (first term) and the Seebeck-driven AHE: the STTG contribution (second term). Importantly, the second term can be designed by the combination of the thermoelectric and magnetic layers as well as their dimensions, i.e., rr. With increasing rr, the second term becomes effective and approaches tanθH(SMSTE)\tan\theta_{\text{H}}(S_{\text{M}}-S_{\text{TE}}), where θHtan1(ρAHE/ρM)\theta_{\text{H}}\equiv\tan^{-1}(\rho_{\text{AHE}}/\rho_{\text{M}}) is the anomalous Hall angle of the magnetic layer. Therefore, a large second term needs large values of θH\theta_{\text{H}}, STES_{\text{TE}}, and rr because STESMS_{\text{TE}}\gg S_{\text{M}} for typical materials.

We demonstrate the behavior of StotyS_{\text{tot}}^{y} using the parameters of typical thermoelectric materials (n-type Si(1), n-type Si(2), p-type Si, and Bi2Te3\text{Bi}_{2}\text{Te}_{3}) and magnetic materials (Ni, Co2MnGa\text{Co}_{2}\text{MnGa}, L10\text{L}1_{0}-ordered FePt, NdFeB, and SmCo5\text{SmCo}_{5}) shown in Table 1, where n-type Si(1) and (2) have different transport properties and hereafter "type" is omitted. Co2MnGa\text{Co}_{2}\text{MnGa} is a Heusler ferromagnet showing large AHE and ANE.Sakai2018NatPhysCo2MnGa ; Reichlova2018APLCo2MnGafilm ; SakurabaCMG , L10\text{L}1_{0}-ordered FePt is often used in spintronics because it has a strong perpendicular magnetic anisotropy in a thin film form. Seki2008NmatFePt1 ; Seki2011APLFePt2 NdFeB and SmCo5\text{SmCo}_{5} are rare-earth permanent magnets in practical use, which are known to exhibit substantially large AHE. Miura2019APLSmCo5 In the previous work by Zhou et al., Ni, Co2MnGa\text{Co}_{2}\text{MnGa}, and FePt were used for the experimental demonstration of STTG. Zou2020 Figure 2(a) [2(b)] shows StotyS_{\text{tot}}^{y} as a function of rr for the combination of the thermoelectric materials and Ni (Co2MnGa\text{Co}_{2}\text{MnGa}). In the case of Ni with negative tanθH\tan\theta_{\text{H}}, the second term of Eq. (5) is negative (positive) for n(p)-Si because of negative (positive) STES_{\text{TE}} and can be larger than SANES_{\text{ANE}} owing to large STES_{\text{TE}} when rr is large. We also find that the rr value at which the Seebeck-driven contribution is apparent with respect to SANES_{\text{ANE}} for n-Si(1) and p-Si is larger than that for n-Si(2) and Bi2Te3\text{Bi}_{2}\text{Te}_{3} due to small ρTE\rho_{\text{TE}} of n-Si(2) and Bi2Te3\text{Bi}_{2}\text{Te}_{3}. As shown in Fig. 2(b), the second term of Eq. (5) for Co2MnGa\text{Co}_{2}\text{MnGa} with positive tanθH\tan\theta_{\text{H}} has the opposite sign to that for Ni for the same thermoelectric material. Importantly, when the magnetic material with large tanθH\tan\theta_{\text{H}} is used, we can obtain the transverse thermopower of >100μK1>100\ \mu\text{V }\text{K}^{-1}, which is more than an order of magnitude larger than SANES_{\text{ANE}} of Co2MnGa\text{Co}_{2}\text{MnGa}. The STTG contribution can further be increased in proportion to tanθH\tan\theta_{\text{H}}, as exemplified in the inset to Fig. 2(b). In Figs. 2(c)-2(e), we show the behavior of StotyS_{\text{tot}}^{y} as a function of rr for L10\text{L}1_{0}-ordered FePt [Fig. 2(c)], NdFeB [Fig. 2(d)], and SmCo5\text{SmCo}_{5} [Fig. 2(e)], which enable STTG in the absence of external magnetic fields owing to their large coercive force and remanent magnetization.Miura2019APLSmCo5 ; Zou2020 We find that NdFeB with relatively large tanθH\tan\theta_{\text{H}} exhibits a larger STTG contribution than FePt and SmCo5\text{SmCo}_{5}, although |SANE||S_{\text{ANE}}| of NdFeB is smaller than that of FePt and SmCo5\text{SmCo}_{5}.Miura2019APLSmCo5 These demonstrations show that the transverse thermopower induced by STTG can be more than an order of magnitude larger than that induced by ANE by optimizing the combination of the thermoelectric and magnetic materials as well as their dimensions.

Here, we show that the insulator layer in Fig. 1 is important to obtain large STTG. For the system in which thermoelectric and magnetic materials are directly connected in the zz direction, we can derive the transverse thermopower by solving Eq. (1) with the boundary condition based on the Maxwell’s equations: ETEx(y)=EMx(y)E_{\text{TE}}^{x(y)}=E_{\text{M}}^{x(y)}. When the closed circuit is formed in the xx direction, the equivalent circuit in the yy direction gives VMy=(IMy+ITEy)RloadV_{\text{M}}^{y}=(I_{\text{M}}^{y}+I_{\text{TE}}^{y})R_{\text{load}}. The transverse thermopower in this case is calculated as

Sshunty=11+rρeffyρTEStoty,S_{\text{shunt}}^{y}=\frac{1}{1+r\frac{\rho_{\text{eff}}^{y}}{\rho_{\text{TE}}}}S_{\text{tot}}^{y}, (6)

where we assume LTEx(y)=LMx(y)L_{\text{TE}}^{x(y)}=L_{\text{M}}^{x(y)} and the linear temperature gradient along the xx direction for simplicity. Here, ρeffy\rho_{\text{eff}}^{y} is the effective transverse resistivity of the STTG system Zou2020 :

ρeffyρM+ρAHE2ρTE/r+ρM.\rho_{\text{eff}}^{y}\equiv\rho_{\text{M}}+\frac{\rho_{\text{AHE}}^{2}}{\rho_{\text{TE}}/r+\rho_{\text{M}}}. (7)

Since the shunting factor 1/[1+r(ρeffy/ρTE)]11/[1+r(\rho_{\text{eff}}^{y}/\rho_{\text{TE}})]\leq 1, we obtain smaller transverse thermopower when the thermoelectric and magnetic materials are directly connected: SshuntyStotyS_{\text{shunt}}^{y}\leq S_{\text{tot}}^{y}. In the following, therefore, we focus only on the case shown in Fig. 1.

III.2 Power factor

We next discuss the power factor (PF) for the STTG system. To derive PF,Zou2020 we first calculate the maximum output power generated by the transverse thermoelectric voltage, Pout(VMy)2/RloadP_{\text{out}}\equiv(V_{\text{M}}^{y})^{2}/R_{\text{load}}, with respect to the load resistance. We calculate the maximum power (Pout)max(P_{\text{out}})_{\text{max}} as follows:

(Pout)max=LMyLMzLMx(StotyΔT)24ρeffyLMxLMyLMz(StotyxT)24ρeffy,(P_{\text{out}})_{\text{max}}=\frac{L_{\text{M}}^{y}L_{\text{M}}^{z}}{L_{\text{M}}^{x}}\frac{({S_{\text{tot}}^{y}}\Delta T)^{2}}{4\rho_{\text{eff}}^{y}}\simeq L_{\text{M}}^{x}L_{\text{M}}^{y}L_{\text{M}}^{z}\frac{({S_{\text{tot}}^{y}}\nabla_{x}T)^{2}}{4\rho_{\text{eff}}^{y}}, (8)

where we used ΔTLxxT\Delta T\simeq-L^{x}\nabla_{x}T with xTxTTExTM\nabla_{x}T\simeq\nabla_{x}T_{\text{TE}}\simeq\nabla_{x}T_{\text{M}}. We then normalize (Pout)max(P_{\text{out}})_{\text{max}} by the temperature gradient and the volume of the magnetic material and obtain PF for the STTG system as

PF(Stoty)2ρeffy.\text{PF}\equiv\frac{{(S_{\text{tot}}^{y})}^{2}}{\rho_{\text{eff}}^{y}}. (9)

This expression becomes equivalent to the power factor for ANE, SANE2/ρM{S_{\text{ANE}}}^{2}/\rho_{\text{M}}, by taking the limit r0r\to 0. Since the second term of Eq. (7) is usually small, the parameter dependence of PF is determined mainly by StotyS_{\text{tot}}^{y}. Although PF in Eq. (9) can be one or two orders of magnitude larger than that for ANE with appropriate choice and design of materials in a similar manner to StotyS_{\text{tot}}^{y}, the difference between the volume of the magnetic material and the total volume of the STTG system should be taken into account to discuss the thermoelectric performance and efficiency of STTG.

Table 1: Parameters of the thermoelectric and magnetic materials.
Thermoelectric materialsQiao2019ACSBe2Te3 ; Zou2020 STE(μK1)S_{\text{TE}}\ (\mu\text{V }\text{K}^{-1}) κTE(m1K1)\kappa_{\text{TE}}\ (\text{W }\text{m}^{-1}\ \text{K}^{-1}) ρTE(Ωm)\rho_{\text{TE}}\ (\Omega\ \text{m})
n-Si(1) 1.3×103-1.3\times 10^{3} 147147 4.1×1024.1\times 10^{-2}
n-Si(2) 6.6×102-6.6\times 10^{2} 121121 1.1×1041.1\times 10^{-4}
p-Si 1.3×1031.3\times 10^{3} 157157 9.9×1029.9\times 10^{-2}
Bi2Te3\text{Bi}_{2}\text{Te}_{3} 1.5×102-1.5\times 10^{2} 11 1.3×1051.3\times 10^{-5}
Magnetic materialsMiura2019APLSmCo5 ; Miura2020PRM ; Sakai2018NatPhysCo2MnGa ; Zou2020 SM(μK1)S_{\text{M}}\ (\mu\text{V }\text{K}^{-1}) κM(Wm1K1)\kappa_{\text{M}}\ (\text{W}\ \text{m}^{-1}\ \text{K}^{-1}) ρM(Ωm)\rho_{\text{M}}\ (\Omega\ \text{m}) ρAHE(Ωm)\rho_{\text{AHE}}\ (\Omega\ \text{m}) tanθH\tan\theta_{\text{H}} SANE(μK1)S_{\text{ANE}}\ (\mu\text{V }\text{K}^{-1})
Ni 21.0-21.0 7676 8.9×1088.9\times 10^{-8} 4.5×1010-4.5\times 10^{-10} 0.005-0.005 0.10.1
Co2MnGa\text{Co}_{2}\text{MnGa} 38.7-38.7 2222 2.7×1062.7\times 10^{-6} 2.4×1072.4\times 10^{-7} 0.0890.089 6.36.3
FePt 21.3-21.3 3030 7.9×1077.9\times 10^{-7} 3.5×1083.5\times 10^{-8} 0.0440.044 1.71.7
NdFeB 5.7-5.7 1010 1.4×1061.4\times 10^{-6} 9.8×108-9.8\times 10^{-8} 0.071-0.071 0.8-0.8
SmCo5\text{Sm}\text{Co}_{5} 18.9-18.9 1414 5.5×1075.5\times 10^{-7} 1.7×108-1.7\times 10^{-8} 0.030-0.030 3.13.1

III.3 Figure of merit

Now, we are in a position to discuss the figure of merit for the transverse thermoelectric conversion in the STTG system. We maximize the efficiency ηPout/(QTEx|T=Th+QMx|T=Th)\eta\equiv P_{\text{out}}/(Q_{\text{TE}}^{x}|_{T=T_{\text{h}}}+Q_{\text{M}}^{x}|_{T=T_{\text{h}}}) with respect to the load resistance RloadR_{\text{load}}, where QTE(M)x|T=ThQ_{\text{TE(M)}}^{x}|_{T=T_{\text{h}}} is the heat current from the hot reservoir in the thermoelectric (magnetic) material in the xx direction. We then obtain the maximum efficiency as

ηmax=ηc11ZtotT¯1+TcTh1ZtotT¯,\eta_{\text{max}}=\eta_{\text{c}}\frac{1-\sqrt{1-Z_{\text{tot}}\overline{T}}}{1+\frac{T_{\text{c}}}{T_{\text{h}}}\sqrt{1-Z_{\text{tot}}\overline{T}}}, (10)

where ηc=1Tc/Th\eta_{\text{c}}=1-T_{\text{c}}/T_{\text{h}} is the Carnot efficiency and T¯=(Th+Tc)/2\overline{T}=(T_{\text{h}}+T_{\text{c}})/2 is the average temperature of the hot and cold reservoirs. Here

ZtotT¯(Stoty)2ρeffyκeffxT¯Z_{\text{tot}}\overline{T}\equiv\frac{(S_{\text{tot}}^{y})^{2}}{\rho_{\text{eff}}^{y}\kappa_{\text{eff}}^{x}}\overline{T} (11)

is the isothermal figure of merit for the STTG system, where

κeffxκM+rκTE+T¯(SMSTE)2ρTE/r+ρM\kappa_{\text{eff}}^{x}\equiv\kappa_{\text{M}}+r\kappa_{\text{TE}}+\frac{\overline{T}(S_{\text{M}}-S_{\text{TE}})^{2}}{\rho_{\text{TE}}/r+\rho_{\text{M}}} (12)

is the effective thermal conductivity of the STTG system with κTE(M)\kappa_{\text{TE(M)}} being the thermal conductivity of the thermoelectric (magnetic) material. Here, ZtotT¯1Z_{\text{tot}}\overline{T}\leq 1, which is a characteristic of the isothermal figure of merit for the transverse thermoelectric conversion.Harman1962JAPTheoryNernst In the same manner as StotyS_{\text{tot}}^{y} and PF, ZtotT¯Z_{\text{tot}}\overline{T} reduces to the isothermal figure of merit for ANE Harman1962JAPTheoryNernst when r0r\to 0.

Although ZtotT¯Z_{\text{tot}}\overline{T} is a complicated function of parameters of thermoelectric and magnetic materials, ZtotT¯Z_{\text{tot}}\overline{T} can be enhanced for a combination of a magnetic material with large tanθH\tan\theta_{\text{H}} and a thermoelectric material with large figure of merit for the Seebeck effect, ZTET¯=STE2T¯/(ρTEκTE)Z_{\text{TE}}\overline{T}={S_{\text{TE}}}^{2}\overline{T}/(\rho_{\text{TE}}\kappa_{\text{TE}}), as discussed in Ref. Zou2020, . This can be intuitively understood with naive approximations as follows. The second term of ρeffy\rho_{\text{eff}}^{y} [Eq. (7)] and the third term of κeffx\kappa_{\text{eff}}^{x} [Eq. (12)] are negligibly small for usual materials. In addition, we assume ρMκMρTEκTE\rho_{\text{M}}\kappa_{\text{M}}\ll\rho_{\text{TE}}\kappa_{\text{TE}} and consider only the STTG contribution in StotyS_{\text{tot}}^{y}. With these approximations, we find that the figure of merit for STTG takes its maximum of ZtotT¯=ZTET¯(tanθH)2/4Z_{\text{tot}}\overline{T}=Z_{\text{TE}}\overline{T}(\tan\theta_{\text{H}})^{2}/4 at r=ρTE/ρMr=\rho_{\text{TE}}/\rho_{\text{M}}. This approximation means that a maximum value of the figure of merit for STTG is mainly determined by tanθH\tan\theta_{\text{H}} and ZTET¯Z_{\text{TE}}\overline{T}.

Refer to caption
Figure 3: The figure of merit ZtotT¯Z_{\text{tot}}\overline{T} [Eq. (11)] for the STTG system at T¯=300 K\overline{T}=300\text{ K} as a function of rr and the Seebeck coefficient of the thermoelectric material STES_{\text{TE}}. In (a), the transport parameters of Co2MnGa\text{Co}_{2}\text{MnGa} are assumed. In (b), tanθH=0.5\tan\theta_{\text{H}}=0.5 is assumed, where the transport parameters except for ρAHE\rho_{\text{AHE}} are fixed at the values for Co2MnGa\text{Co}_{2}\text{MnGa}. Although STES_{\text{TE}} is treated as a parameter, STE2/ρTE{S_{\text{TE}}}^{2}/\rho_{\text{TE}} and κTE\kappa_{\text{TE}} are fixed at the values for Bi2Te3\text{Bi}_{2}\text{Te}_{3}, i.e., ZTET¯=const.Z_{\text{TE}}\overline{T}=\text{const.}, where the black curve in (a) shows the case for Co2MnGa/Bi2Te3\text{Co}_{2}\text{MnGa}/\text{Bi}_{2}\text{Te}_{3} and the plane surrounded by dotted black lines shows the figure of merit for ANE in Co2MnGa\text{Co}_{2}\text{MnGa}. The red (blue) color shows the larger (smaller) figure of merit than that for ANE in Co2MnGa\text{Co}_{2}\text{MnGa}.

The behaviors of ZtotT¯Z_{\text{tot}}\overline{T} are exemplified in Fig. 3 (note that the aforementioned approximations are not applied here). Figure 3(a) shows ZtotT¯Z_{\text{tot}}\overline{T} for Co2MnGa\text{Co}_{2}\text{MnGa} at T¯=300 K\overline{T}=300\text{ K} as a function of rr and STES_{\text{TE}}. Although STES_{\text{TE}} is treated as a parameter, STE2/ρTE{S_{\text{TE}}}^{2}/\rho_{\text{TE}} and κTE\kappa_{\text{TE}} are fixed at the values for Bi2Te3\text{Bi}_{2}\text{Te}_{3}, i.e., ZTET¯=const.Z_{\text{TE}}\overline{T}=\text{const.}, where the black curve in Fig. 3(a) shows the case for the Co2MnGa\text{Co}_{2}\text{MnGa}/Bi2Te3\text{Bi}_{2}\text{Te}_{3} junction with the parameters shown in Table 1. By choosing optimal rr, ZtotT¯Z_{\text{tot}}\overline{T} can be larger than the figure of merit for ANE in Co2MnGa\text{Co}_{2}\text{MnGa} both for positive and negative STES_{\text{TE}} values. The difference in the maximum ZtotT¯Z_{\text{tot}}\overline{T} between positive and negative STES_{\text{TE}} values is attributed to the SANES_{\text{ANE}} offset in Eq. (5); the situation depends on the sign and magnitude of SANES_{\text{ANE}}. As shown in Fig. 3(b), when tanθH=0.5\tan\theta_{\text{H}}=0.5 is assumed, the maximum value of ZtotT¯Z_{\text{tot}}\overline{T} is dramatically improved in comparison with the figure of merit for ANE in Co2MnGa\text{Co}_{2}\text{MnGa}. Here, due to the dominant contribution of STTG, the significant improvement of ZtotT¯Z_{\text{tot}}\overline{T} appears both for positive and negative STES_{\text{TE}} values. This demonstration shows that not only StotyS_{\text{tot}}^{y} and PF but also ZtotT¯Z_{\text{tot}}\overline{T} can be enhanced by AHE driven by the Seebeck effect.

III.4 Thermoelectric cooling

Finally, we mention that the STTG system also works as a transverse thermoelectric temperature modulator by replacing the load resistance with an external battery in Fig. 1. The external battery induces a charge current in the yy direction, and the charge current is bent in the xx direction by AHE. When the charge current in the xx direction flows in the closed circuit comprising the thermoelectric and magnetic materials, heat is generated or absorbed at the junctions by the Peltier effect. This is the reciprocal process of STTG. The resultant temperature gradient under the adiabatic condition in the xx direction is calculated as follows. Since we use the same boundary conditions for TTE(M)T_{\text{TE(M)}}, the temperature gradients in the thermoelectric and magnetic materials are the same as Eqs. (3) and (4), respectively. By solving Eq. (1) with the boundary conditions ITEx+IMx=0I_{\text{TE}}^{x}+I_{\text{M}}^{x}=0, VTEx=VMxV_{\text{TE}}^{x}=V_{\text{M}}^{x}, we obtain IMxI_{\text{M}}^{x} and IMyI_{\text{M}}^{y} as follows:

IMx\displaystyle I^{x}_{\text{M}} =SMSTERTEx+RMxΔT+LMxLMyRAHERTEx+RMxIMy,\displaystyle=\frac{S_{\text{M}}-S_{\text{TE}}}{R_{\text{TE}}^{x}+R_{\text{M}}^{x}}\Delta T+\frac{L_{\text{M}}^{x}}{L_{\text{M}}^{y}}\frac{R_{\text{AHE}}}{R_{\text{TE}}^{x}+R_{\text{M}}^{x}}I_{\text{M}}^{y}, (13)
IMy\displaystyle I_{\text{M}}^{y} =VMyReffyLMyLMxStotyReffyΔT,\displaystyle=-\frac{V_{\text{M}}^{y}}{R_{\text{eff}}^{y}}-\frac{L_{\text{M}}^{y}}{L_{\text{M}}^{x}}\frac{S_{\text{tot}}^{y}}{R_{\text{eff}}^{y}}\Delta T, (14)

where RTE(M)x=ρTE(M)LTE(M)x/(LTE(M)yLTE(M)z)R_{\text{TE(M)}}^{x}=\rho_{\text{TE(M)}}L_{\text{TE(M)}}^{x}/(L_{\text{TE(M)}}^{y}L_{\text{TE(M)}}^{z}) is the resistance of the thermoelectric (magnetic) material in the xx direction, RAHE=ρAHELMy/(LMzLMx)R_{\text{AHE}}=\rho_{\text{AHE}}L_{\text{M}}^{y}/(L_{\text{M}}^{z}L_{\text{M}}^{x}) is the anomalous Hall resistance, and Reffy=ρeffyLMy/(LMzLMx)R_{\text{eff}}^{y}=\rho_{\text{eff}}^{y}L_{\text{M}}^{y}/(L_{\text{M}}^{z}L_{\text{M}}^{x}) is the effective resistance in the magnetic material in the yy direction. Using Eq. (13) and QTEx+QMx=0Q_{\text{TE}}^{x}+Q_{\text{M}}^{x}=0, we obtain the resultant temperature gradient under the adiabatic condition in the xx direction as follows:

xT=ΠtotyEMyκeffxρeffy(1ZtotT)ΠtotyEMyκeffxρeffy,\nabla_{x}T=\frac{\Pi_{\text{tot}}^{y}E_{\text{M}}^{y}}{\kappa_{\text{eff}}^{x}\rho_{\text{eff}}^{y}(1-Z_{\text{tot}}T)}\simeq\frac{\Pi_{\text{tot}}^{y}E_{\text{M}}^{y}}{\kappa_{\text{eff}}^{x}\rho_{\text{eff}}^{y}}, (15)

where we assumed ZtotT1Z_{\text{tot}}T\ll 1 and the linear temperature gradient in the xx direction for simplicity. Here, Πtoty\Pi_{\text{tot}}^{y} is the transverse charge-to-heat conversion coefficient for the STTG system defined as

Πtoty=(qMx+rqTExjMy)xT=0,\Pi_{\text{tot}}^{y}=\left(\frac{q_{\text{M}}^{x}+rq_{\text{TE}}^{x}}{j_{\text{M}}^{y}}\right)_{\nabla_{x}T=0}, (16)

where qTE(M)xq_{\text{TE(M)}}^{x} is the heat current density in the thermoelectric (magnetic) material in the xx direction. We note that, in the definition of Πtoty\Pi_{\text{tot}}^{y}, the aforementioned adiabatic condition in the xx direction is not assumed. The coefficient satisfies the reciprocal relation Πtoty=TStoty\Pi_{\text{tot}}^{y}=TS_{\text{tot}}^{y} (see the right longitudinal axis in Fig. 2). By taking the limit r0r\to 0, Πtoty\Pi_{\text{tot}}^{y} reduces to the anomalous Ettingshausen coefficient, ΠAEE=TSANE\Pi_{\text{AEE}}=TS_{\text{ANE}}, and the temperature gradient in Eq. (15) reduces to that for the anomalous Ettingshausen effect.Seki2018JphysDFePtfilm ; Seki2018APL

Refer to caption
Figure 4: (a) The achievable ΔT/T¯\Delta T/\overline{T} [Eq. (18)] as a function of ZtotT¯Z_{\text{tot}}\overline{T}. The black circle shows the achievable ΔT/T¯\Delta T/\overline{T} for the figure of merit for ANE in Co2MnGa\text{Co}_{2}\text{MnGa}. (b) COP [Eq. (17)] normalized by COPlimit\text{COP}_{\text{limit}} and the cooling power density for Co2MnGa/Bi2Te3\text{Co}_{2}\text{MnGa}/\text{Bi}_{2}\text{Te}_{3} at T¯=300 K\overline{T}=300\text{ K} as a function of |IMy||I_{\text{M}}^{y}|, where we set LTE(M)x=LTE(M)y=10 mmL_{\text{TE(M)}}^{x}=L_{\text{TE(M)}}^{y}=10\text{ mm}, LMz=0.1 mmL_{\text{M}}^{z}=0.1\text{ mm}, and LTEz=1 mmL_{\text{TE}}^{z}=1\text{ mm}. (c) The maximum COP [Eq. (19)] normalized by COPlimit\text{COP}_{\text{limit}} at T¯=300 K\overline{T}=300\text{ K} and ΔT=0.05 K\Delta T=0.05\text{ K} as a function of rr and the Peltier coefficient of the thermoelectric material ΠTE\Pi_{\text{TE}}, where the parameters are set in the same manner as Fig. 3(a). The black curve in (c) shows the case for Co2MnGa/Bi2Te3\text{Co}_{2}\text{MnGa}/\text{Bi}_{2}\text{Te}_{3}. We only plot the region where ΔT=0.05 K\Delta T=0.05\text{ K} is achievable.

We define the coefficient of performance (COP) for the STTG system as

COP =QTEx|T=Tc+QMx|T=TcIMyVext\displaystyle=-\frac{Q_{\text{TE}}^{x}|_{T=T_{\text{c}}}+Q_{\text{M}}^{x}|_{T=T_{\text{c}}}}{I_{\text{M}}^{y}V_{\text{ext}}}
=Reffy2(IMy)2+LMyLMxThStotyIMy+KeffxΔTReffy(IMy)2+LMyLMxStotyΔTIMy,\displaystyle=-\frac{\frac{R_{\text{eff}}^{y}}{2}(I_{\text{M}}^{y})^{2}+\frac{L_{\text{M}}^{y}}{L_{\text{M}}^{x}}T_{\text{h}}S_{\text{tot}}^{y}I_{\text{M}}^{y}+K_{\text{eff}}^{x}\Delta T}{R_{\text{eff}}^{y}(I_{\text{M}}^{y})^{2}+\frac{L_{\text{M}}^{y}}{L_{\text{M}}^{x}}S_{\text{tot}}^{y}\Delta TI_{\text{M}}^{y}}, (17)

where Vext=VMyV_{\text{ext}}=-V_{\text{M}}^{y} is the voltage of the external battery and Keffx=κeffxLMyLMz/LMxK_{\text{eff}}^{x}=\kappa_{\text{eff}}^{x}L_{\text{M}}^{y}L_{\text{M}}^{z}/L_{\text{M}}^{x} is the effective thermal conductance of the STTG device. Here, we used Eqs. (2)-(4) and (13) for the numerator, and Eq. (14) for the denominator. To cool the STTG system, the heat current should be absorbed from the cold reservoir, QTEx|T=Tc+QMx|T=Tc0Q_{\text{TE}}^{x}|_{T=T_{\text{c}}}+Q_{\text{M}}^{x}|_{T=T_{\text{c}}}\leq 0, and this gives the following condition for the temperature difference:

ΔTT¯2ZtotT¯(2ZtotT¯21ZtotT¯).\frac{\Delta T}{\overline{T}}\leq\frac{2}{Z_{\text{tot}}\overline{T}}\left(2-Z_{\text{tot}}\overline{T}-2\sqrt{1-Z_{\text{tot}}\overline{T}}\right). (18)

We show the upper bound of ΔT/T¯\Delta T/\overline{T} in Fig. 4(a); this is the achievable ΔT/T¯\Delta T/\overline{T} for a given ZtotT¯Z_{\text{tot}}\overline{T}. Figure 4(b) shows the |IMy||I_{\text{M}}^{y}| dependence of COP and the cooling power density defined as (QTEx|T=Tc+QMx|T=Tc)/LMyLMz-(Q_{\text{TE}}^{x}|_{T=T_{\text{c}}}+Q_{\text{M}}^{x}|_{T=T_{\text{c}}})/L_{\text{M}}^{y}L_{\text{M}}^{z}. The behaviors are similar to those for the conventional Peltier and Ettingshausen effects.Mobarak2021PRAppl The maximum COP with respect to IMyI_{\text{M}}^{y} is calculated to be

COPmax=COPlimit1ThTc1ZtotT¯1+1ZtotT¯,\text{COP}_{\text{max}}=\text{COP}_{\text{limit}}\frac{1-\frac{T_{\text{h}}}{T_{\text{c}}}\sqrt{1-Z_{\text{tot}}\overline{T}}}{1+\sqrt{1-Z_{\text{tot}}\overline{T}}}, (19)

where COPlimit=Tc/ΔT\text{COP}_{\text{limit}}=T_{\text{c}}/\Delta T is the achievable limit of COP. Equation (19) reduces to the maximum COP for the anomalous Ettingshausen effectHarman1962JAPCOP when r0r\to 0. COPmax\text{COP}_{\text{max}} for the STTG system can be larger than that for the anomalous Ettingshausen effect alone; its behavior is similar to that of ZtotT¯Z_{\text{tot}}\overline{T} [compare Fig. 4(c) with 3(a)].

IV Conclusion

We have phenomenologically calculated the thermoelectric properties of STTG for various combinations of thermoelectric and magnetic materials and demonstrated its usefulness. We have shown that, by combining a thermoelectric material with a large Seebeck coefficient and a magnetic material with a large anomalous Hall angle, not only the transverse thermopower but also the figure of merit for STTG can be much larger than those for conventional ANE. We have also discussed the reciprocal process of STTG; the combination of AHE in a magnetic material and the Peltier effect in a thermoelectric material enables transverse charge-to-heat conversion. It is worth mentioning that STTG works even if the thermoelectric material is also ferromagnetic, although the formulation becomes complicated. Thus, the experimental demonstration of STTG in all-ferromagnetic systems is desired. The phenomenological model for STTG will invigorate materials science research and device engineering in spin caloritronics, paving the way for versatile energy harvesting, heat sensing, and thermal management applications.

Acknowledgements.
This work was supported by CREST “Creation of Innovative Core Technologies for Nano-enabled Thermal Management” (Grant No. JPMJCR17I1) and PRESTO “Scientific Innovation for Energy Harvesting Technology” (Grant No. JPMJPR17R5) from JST, Japan, and Mitou challenge 2050 (Grant No. P14004) from NEDO, Japan. A. M. was supported by JSPS through Research Fellowship for Young Scientists (Grant No. JP18J02115).

Data Availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  • (1) G. E. W. Bauer, E. Saitoh, and B. J. Van Wees, Nat. Mater. 11, 391 (2012).
  • (2) S. R. Boona, R. C. Myers, and J. P. Heremans, Energy Environ. Sci. 7, 885 (2014).
  • (3) K. Uchida, H. Adachi, T. Kikkawa, A. Kirihara, M. Ishida, S. Yorozu, S. Maekawa, and E. Saitoh, Proc. IEEE 104, 1946 (2016).
  • (4) M. Mizuguchi and S. Nakatsuji, Sci. Tech. Adv. Mater. 20, 262 (2019).
  • (5) K. Uchida, W. Zhou, and Y. Sakuraba, Appl. Phys. Lett. 118, 140504 (2021).
  • (6) Y. Sakuraba, K. Hasegawa, M. Mizuguchi, T. Kubota, S. Mizukami, T. Miyazaki, and K. Takanashi, Appl. Phys. Express 6, 033003 (2013).
  • (7) Y. Sakuraba, Scr. Mater. 111, 29 (2016).
  • (8) Z. Yang, E. A. Codecido, J. Marquez, Y. Zheng, J. P. Heremans, and R. C. Myers, AIP Adv. 7, 095017 (2017).
  • (9) W. Zhou and Y. Sakuraba, Appl. Phys. Express 13, 043001 (2020).
  • (10) M. Mizuguchi, S. Ohata, K. Uchida, E. Saitoh, and K. Takanashi, Appl. Phys. Express 5, 093002 (2012).
  • (11) K. Hasegawa, M. Mizuguchi, Y. Sakuraba, T. Kamada, T. Kojima, T. Kub- ota, S. Mizukami, T. Miyazaki, and K. Takanashi, Appl. Phys. Lett. 106, 252405 (2015).
  • (12) S. Isogami, K. Takanashi, and M. Mizuguchi, Appl. Phys. Express 10, 073005 (2017).
  • (13) T. Seki, R. Iguchi, K. Takanashi, and K. Uchida, J. Phys. D: Appl. Phys. 51, 254001 (2018).
  • (14) H. Nakayama, K. Masuda, J. Wang, A. Miura, K. Uchida, M. Murata, and Y. Sakuraba, Phys. Rev. Mater. 3, 114412 (2019).
  • (15) A. Sakai, S. Minami, T. Koretsune, T. Chen, T. Higo, Y. Wang, T. Nomoto, M. Hirayama, S. Miwa, D. Nishio-Hamane, F. Ishii, R. Arita, and S. Nakat- suji, Nature (London) 581, 53 (2020).
  • (16) A. Sakai, Y. P. Mizuta, A. A. Nugroho, R. Sihombing, T. Koretsune, M.-T. Suzuki, N. Takemori, R. Ishii, D. Nishio-Hamane, R. Arita, P. Goswami, and S. Nakatsuji, Nat. Phys. 14, 1119 (2018).
  • (17) H. Reichlova, R. Schlitz, S. Beckert, P. Swekis, A. Markou, Y.-C. Chen, D. Kriegner, S. Fabretti, G. Hyeon Park, A. Niemann, S. Sudheendra, A. Thomas, K. Nielsch, C. Felser, and S. T. B. Goennenwein, Appl. Phys. Lett. 113, 212405 (2018).
  • (18) G.-H. Park, H. Reichlova, R. Schlitz, M. Lammel, A. Markou, P. Swekis, P. Ritzinger, D. Kriegner, J. Noky, J. Gayles, Y. Sun, C. Felser, K. Nielsch, S. T. B. Goennenwein, and A. Thomas, Phys. Rev. B 101, 060406 (2020).
  • (19) Y. Sakuraba, K. Hyodo, A. Sakuma, and S. Mitani, Phys. Rev. B 101, 134407 (2020).
  • (20) K. Sumida, Y. Sakuraba, K. Masuda, T. Kono, M. Kakoki, K. Goto, W. Zhou, K. Miyamoto, Y. Miura, T. Okuda, and A. Kimura, Commun. Mater. 1, 89 (2020).
  • (21) A. Miura, H. Sepehri-Amin, K. Masuda, H. Tsuchiura, Y. Miura, R. Iguchi, Y. Sakuraba, J. Shiomi, K. Hono, and K. Uchida, Appl. Phys. Lett. 115, 222403 (2019).
  • (22) A. Miura, K. Masuda, T. Hirai, R. Iguchi, T. Seki, Y. Miura, H. Tsuchiura, K. Takanashi, and K. Uchida, Appl. Phys. Lett. 117, 082408 (2020).
  • (23) K. Uchida, T. Kikkawa, T. Seki, T. Oyake, J. Shiomi, Z. Qiu, K. Takanashi, and E. Saitoh, Phys. Rev. B 92, 094414 (2015).
  • (24) C. Fang, C. H. Wan, Z. H. Yuan, L. Huang, X. Zhang, H. Wu, Q. T. Zhang, and X. F. Han, Phys. Rev. B 93, 054420 (2016).
  • (25) R. Ramos, T. Kikkawa, A. Anadón, I. Lucas, T. Niizeki, K. Uchida, P. A. Algarabel, L. Morellón, M. H. Aguirre, M. R. Ibarra, and E. Saitoh, Appl. Phys. Lett. 114, 113902 (2019).
  • (26) T. Seki, Y. Sakuraba, K. Masuda, A. Miura, M. Tsujikawa, K. Uchida, T. Kubota, Y. Miura, M. Shirai, and K. Takanashi, Phys. Rev. B 103, L020402 (2021).
  • (27) W. Zhou, K. Yamamoto, A. Miura, R. Iguchi, Y. Miura, K. Uchida, and Y. Sakuraba, Nat. Mater. 20, 463 (2021).
  • (28) T. C. Harman and J. M. Honig, J. Appl. Phys. 33, 3178 (1962).
  • (29) L. D. Landau, E. Lifshitz, and L. Pitaevskii, Electrodynamics of continuous media (Pergamon Press, Oxford, 1984).
  • (30) Y. Zhang, N. P. Ong, Z. A. Xu, K. Krishana, R. Gagnon, and L. Taillefer, Phys. Rev. Lett. 84, 2219 (2000).
  • (31) X. Li, L. Xu, L. Ding, J. Wang, M. Shen, X. Lu, Z. Zhu, and K. Behnia, Phys. Rev. Lett. 119, 056601 (2017).
  • (32) T. Seki, Y. Hasegawa, S. Mitani, S. Takahashi, H. Imamura, S. Maekawa, J. Nitta, and K. Takanashi, Nat. Mater. 7, 125 (2008).
  • (33) T. Seki, M. Kohda, J. Nitta, and K. Takanashi, Appl. Phys. Lett. 98, 212505 (2011).
  • (34) J. Qiao, Y. Zhao, Q. Jin, J. Tan, S. Kang, J. Qiu, and K. Tai, ACS Appl. Mater. Interfaces 11, 38075 (2019).
  • (35) A. Miura, R. Iguchi, T. Seki, K. Takanashi, and K. Uchida, Phys. Rev. Mater. 4, 034409 (2020).
  • (36) T. Seki, R. Iguchi, K. Takanashi, and K. Uchida, Appl. Phys. Lett. 112, 152403 (2018).
  • (37) M. M. H. Polash and D. Vashaee, Phys. Rev. Appl. 15, 014011 (2021).
  • (38) T. C. Harman and J. M. Honig, J. Appl. Phys. 33, 3188 (1962).