WorksPhysics and Astronomy
High Thermoelectric Cooling Performance of Junction Thermoelectric Transistors
Abstract
To achieve high performance thermoelectric materials and devices, thermoelectric transistors, which integrate thermoelectric effects with transistor technology, represent a promising approach. Here p type Bi0.5Sb1.5Te3 and n type Bi2Te2.97Se0.03 are used as the constituent materials for an NPN transistor. By applying forward bias to the emitter and reverse bias to the collector to form a common-base triode configuration, the thermoelectric effect, transistor effect, and interfacial effects within the NPN heterostructure are coupled. This NPN heterostructure induces temperature increase (heat release) at the forward biased end and temperature decrease (heat absorption) at the reverse biased end. Therefore, this device becomes a new type of thermoelectric transistor cooler. Furthermore, a DC equivalent circuit method is introduced to analyze the cooling performance of the thermoelectric transistor cooler. The results show that when the emitter voltage of the NPN thermoelectric transistor cooler is 0.01925V, the collector voltage is 0.2124V, the corresponding base region width is 0.17nm, the maximum temperature difference of 242.89K can be obtained. Even if the base region width is limited to a level of 10 nanometers, for example when the base region width is 12.78nm, the maximum temperature differenced can still reach 174.15K. The research results above indicate that the thermoelectric transistor, which synergistically combines the thermoelectric effect with transistor technology, can effectively enhance the maximum temperature difference that typical thermoelectric cooling can achieve.
Introduction
The continuous demand for energy-efficient cooling solutions, combined with environmental concerns, highlights the importance of exploring innovative methods to reduce global energy consumption and greenhouse gas emissions. Therefore, thermoelectric (TE) cooling has become one of the preferred solutions that attracts much attention.
Thermoelectric cooling, also called semiconductor refrigeration or thermoelectric refrigeration, is a new refrigeration technology developed in the late 1950s. It achieves refrigeration through the Peltier effect by directly using electrical energy. The Peltier effect was discovered by J. Peltier [1] in 1834, referring to the phenomenon that when electric current flows through the junction of two different materials, heat is absorbed from or released into the external environment. In practical applications, thermoelectric cooling possesses the advantages of: fast response, no pollution, no noise, no moving parts, small size, light weight, strong adaptability (adjustable in size, shape, temperature, power, etc.), high reliability (no wear, basically no maintenance required), high mechanical strength, strong impact resistance, long service life, coefficient of performance not depending on capacity, easy speed control, reversible operation [2, 3]. As a result, thermoelectric cooling has been widely applied.
Despite having significant advantages in specific application domains such as aerospace, electronic devices, and healthcare, thermoelectric cooling technology is still constrained for large-scale commercialization by its relatively low conversion efficiency [4]. In recent years, scholars have pursued three primary strategies to enhance the performance of thermoelectric coolers:
First, enhancing the thermoelectric properties of materials [5, 6]. The performance of thermoelectric materials is principally determined by the dimensionless figure of merit , where , , , and denote the Seebeck coefficient, electrical conductivity, absolute temperature, and total thermal conductivity (comprising lattice contribution and electronic contribution ) respectively [7, 8, 9, 10, 11, 12]. Material optimization approaches include alloying [13, 14, 15], synthesis of nanocomposite structures [16, 17, 18, 19], low-dimensional engineering [20, 21, 22, 23, 24]. Second, geometric and structural optimization of thermoelectric devices. Currently, the mainstream thermoelectric cooling device structures are three types: unipolar/layered devices constructed from series/parallel-connected -type thermoelements (fundamental N/P-type material units), multistage thermoelectric coolers, Y-shaped structural devices [25]. Third, exploring novel concepts and operational principles – the primary focus of this research.
The combination of thermoelectric principles and transistor technology has attracted widespread attention, including electric-field-driven thermoelectric devices [26, 27, 28, 29, 30] and purely temperature-gradient-driven junction thermoelectric transistors [31, 32, 33]. Research indicates that in electric-field-driven thermoelectric devices, under the combined action of temperature gradient and gate voltage, thermoelectric performance can be optimized by independently regulating internal material parameters and decoupling the Seebeck coefficient and electrical conductivity [26]. Bejenari et al. used theoretical models to study the influence of gate voltage on the thermoelectric properties of BiTe nanowires [27]. The results indicated that the of 7nm-thick BiTe nanowires could reach up to 3.4. Subsequently, experimental results from Qin et al. showed that under gate voltage, the values of n-type and p-type BiTe thin films could achieve 1.22 and 1.02, respectively [28]. Through experimental measurements, Yang et al. discovered that the power factor () of 2D-BiOSe could exceed 400 WmK due to gate voltage effects [29]. Additionally, Wu et al. found that under gate voltage, 2D NbSiTe-based transistors realized the Seebeck diode effect at room temperature [30].
Thermoelectric junction transistors based on PNP/NPN heterostructures demonstrate substantial potential in thermoelectric by overcoming the inherent limitations of strong parameter coupling in conventional systems. With advancing applications of P-N junctions in thermoelectricity, research confirms that interfacial effects at P-N junctions enable significantly enhanced Seebeck coefficients near the depletion layer [31, 32, 33, 34, 35, 36]. In 1986, Balmush et al. postulated that voltages across P-N junctions under non-isothermal conditions could increase dramatically near depletion regions [37]. Subsequent theoretical work in the 1990s by Zakhidov, Ravich, and collaborators demonstrated through detailed calculations that P-N junctions could substantially elevate both Seebeck coefficient and values via interfacial phenomena [38, 39, 40]. Nan et al. systematically investigated the thermoelectric conversion performance of -based PNP junction thermoelectric transistors [31, 32, 33]. Through optimizing the carrier concentrations of the emitter, base, and collector, as well as the device dimensions, and utilizing thermal equilibrium equations, the PNP thermoelectric junction transistor achieves a maximum output power of 17.8mW under a 50K temperature difference drive, with a corresponding conversion efficiency of 8.69%, under maximum power output mode without considering the Thomson effect [32]. This efficiency would require conventional thermoelectric generators to achieve , underscoring the substantial potential of junction thermoelectric transistors for breakthrough performance enhancement.
Accordingly, in this paper, P-type and N-type are used to construct NPN junction thermoelectric transistors. Positive and negative bias voltages are applied to the emitter and collector respectively to regulate the thermoelectric effect, interface effect and transistor effect of the junction-type thermoelectric transistor, thereby decoupling material parameters and improving the thermoelectric cooling performance. A systematic study is conducted on the output performance of the thermoelectric transistor cooling device.
Principles and Analytical Methods
Working principle and material selection of TE transistor cooling devices
The working principle of the general thermoelectric cooling device or the Peltier effect in semiconductor materials can be explained by Figure 1[41]. As can be seen from Figure 1, when the current flows from the N type region to the P type region, the charge carriers continuously flow out at the pn junction (1). Therefore, new charge carriers need to be generated accordingly, which requires energy consumption, thus causing the temperature to decrease. When the charge carriers flow to the pn junction (2), the two types of charge carriers move towards each other, resulting in the recombination of electrons and holes, thereby releasing energy and causing the temperature to rise. It can also be seen from Figure 1 that the pn junction (1) is reverse-biased, and the pn junction (2) is forward-biased. That is, reverse bias causes the interface temperature to decrease, and forward bias causes the interface temperature to increase. Moreover, the currents passing through the two pn junctions in Figure 1 cannot be independently controlled because they share a single power supply.

Figure 1. Schematic diagram of the Peltier effect of semiconductor materials
Here, the rectangular structure design of the NPN heterojunction is shown in Figure 2 (a). Along the X direction, , , and represent the widths of N-type, P-type, and N-type materials, respectively. In the YZ plane, the cross-sectional area of the NPN heterojunction is A. The P-type region is fully depleted to eliminate the effects of bipolar diffusion. Therefore, there are no free minority carriers (holes) in this P-type region. And the dimensions of and are at the nanoscale. The circuit connection structure diagram of this NPN junction transistor used for refrigeration is shown in Figure 2 (b). The voltages applied between the N-P junction and the N-P junction ensure that the N-P junction is in reverse bias while the N-P junction is in forward bias. When current flows through the circuit, the temperature at the right end of the N region will increase, while the temperature at the left end of the N region will decrease, thereby creating a temperature difference between the two ends of the the NPN junction thermoelectric transistor. Compared with Figure 1, the fundamental difference between the two figures lies in that the voltages of the two pn junctions in Figure 2 can be independently controlled. Therefore, by adjusting their voltages, the temperature of the N-P junction at the high-temperature end in Figure 2 can be lowered, which is more conducive to improving the cooling effect of the semiconductor cooling device.

Figure 2. (a) Schematic diagram of the thermoelectric transistor structure; (b) Circuit configuration of the thermoelectric transistor
The working principle is as follows: By controlling the carrier concentration of emitter N and base P, as well as V, the current through the interface between collector N and base P can be regulated. By adjusting the carrier concentration of N and V, the cooling capacity can be controlled. This is because in a junction transistor, the emitter current J is approximately equal to the collector current J, i.e., J J. Therefore, by optimizing and controlling the carrier concentration of emitter N, base P, and collector N, as well as the applied voltages V and V, the cooling capacity of the collector can be maximized.
In this work, N-BiTeSe, P-BiSbTe, and N-BiTeSe are employed as the collector, base, and emitter materials, respectively, with their essential parameters summarized in Table 1.
Table 1. Material Parameters of TE transistors
Carrier concentration analysis
In thermoelectric transistors, since the P type region is completely depleted, only the charge carriers (electrons) in the N-type region can move through the NP or PN junction. In this work, N-type is used as the N type thermoelectric material; therefore, the majority electrons contribute more to the thermoelectric process than the minority holes [48]. Hereafter, we focus on the electron concentration distribution of N-type . The distribution of electrons can be obtained through a theoretical analysis method. This method consists of three equations: the current density expression for DC current transport [49], the one-dimensional Poisson equation [47], and the current continuity equation [50]:
The current density equation governing DC charge transport can be expressed as formula (1):
The one-dimensional Poisson equation can be expressed as formula (2):
The current continuity equation can be expressed as formula (3):
Where represents the current density, and denote the electrical conductivity and Seebeck coefficient of N-type , signifies the temperature gradient, stands for the electric field, indicates the electric potential, is the charge, is the relative dielectric constant, is the vacuum dielectric constant, is the electron concentration distribution in N-type , represents the acceptor doping concentration of N-type , represents the generation rate of electron carriers, and denotes the recombination rate of electron carriers.
Since the N-type used in this work can be approximated as a degenerate semiconductor, the conductivity and the seebeck coefficient can be expressed as formula (4) [47][51]:
Where is the electron mobility, is the Boltzmann constant, is the electron charge, is the scattering factor, is the effective mass of the carrier, is the Planck constant, and is the temperature distribution.
At the high-temperature end (the right end of N2), a constant temperature boundary condition is adopted. Assuming that the temperature in the N-type is linearly distributed, with the temperature at the high-temperature end (the right end of N2) being and the temperature at the low-temperature end (the left end of N1) being , the temperature distribution can be expressed as:
Since the simulation temperature is lower than the intrinsic excitation temperature of the N-type Bi₂Te₂.₉₇Se₀.₀₃ and it can be assumed that the electron generation rate equals the recombination rate, thus the electron concentration remains unchanged.
Then, formula (1b) becomes:
Substituting equation (7) into equation (1a) yields
Meanwhile, assuming that the temperature of the N-type Bi₂Te₂.₉₇Se₀.₀₃ (including the N₁ and N₂ regions) is in a dynamically stable distribution state, then
Finally, based on equations (4a), (4b), (5), (8), and (9), the electron concentration distribution in N-type Bi₂Te₂.₉₇Se₀.₀₃ can be calculated.
In the thermoelectric transistor of this work, both the N₁ and N₂ regions are composed of N-type Bi₂Te₂.₉₇Se₀.₀₃. Since the P-type region is extremely thin (only a few nanometers in thickness), it is assumed that its temperature remains constant at . The temperature distributions are as follows:
Where and represent the widths of the N₁ and N₂ regions, respectively. Based on equations (4a), (4b), (5), (8), (9), (10), and (11), the electron concentration distributions in the N₁ and N₂ regions are given by:
Thus, the electron concentration in the N₁ region at the N₁-P interface and the electron concentration in the N₂ region at the N₂-P interface are respectively:
Operating States of the TE Transistor
In this thermoelectric transistor, the N₁, P and N₂ regions are used as the collector, base, and emitter, respectively. The applied voltages and divide the thermoelectric transistor into reverse-biased and forward-biased sections.
In the reverse-biased section of the thermoelectric transistor, the P-type region is fully depleted. Consequently, the Seebeck voltage and Seebeck coefficient in the region are given by:
Where is the Boltzmann constant, is the electron charge, is the scattering factor, is the carrier effective mass, and is the Planck constant.
The built-in voltage of the -P junction is given by:
Where and represent the intrinsic carrier concentrations of N-type BiTeSe and P-type BiSbTe, respectively. denotes the electron concentration in the region at temperature , is the hole concentration in the P-type region. Since the P-type region is at the nanoscale, it is assumed that is equal to the doping concentration of the P-type region.
Similarly, the Seebeck voltage and Seebeck coefficient , in the region, and the built-in voltage at the -P interface are given by:
Where represents the Seebeck coefficient in the region, denotes the electron concentration in the region at temperature (which equals its acceptor doping concentration), and is the hole concentration in the P-type region. Since the P-type region is at the nanoscale, it is assumed that is equal to the doping concentration of the P-type region.
Working Principle and DC Equivalent Circuit Diagram of the TE Transistor Cooling Devices
In the thermoelectric transistor circuit, the base-collector leakage current and Early voltage effects are neglected. The equivalent DC circuit is illustrated in Figure 2. The left section represents the reverse-biased part, where is the Seebeck voltage of the -P region, is the total resistance of the -P region including the resistances of both and P regions as well as the interface resistance, and is the interface voltage between and P. The right part corresponds to the forward-biased section, where is the Seebeck voltage generated by the -P region, is the total resistance of the -P region including the resistances of both and P regions as well as the interface resistance, and is the interface voltage between and P. Additionally, the contact resistance between the thermoelectric transistor and electrodes is neglected in this model.
The internal equivalent circuit of the TE transistor is represented within the black dashed box in the Figure 3.

Figure 3. DC equivalent circuit diagram of the thermoelectric transistor cooling device.
According to Kirchhoff's laws, the relationships between these electrical parameters in the thermoelectric transistor can be expressed as:
Additionally
Where represents the current density, denotes the cross-sectional area of the thermoelectric transistor, and are the diffusion lengths of P-type and N-type , respectively, and and are the charge carrier mobilities of P-type and N-type . The diffusion coefficients , , , and correspond to the reverse-biased and forward-biased regions of P-type and N-type , respectively. These diffusion coefficients () can be derived from the Einstein relation:
Since the thermoelectric cooling process belongs to high current injection, the current density formula is given by:
Since P₁ and P₂ are completely depleted, and can be expressed as
Where ε₁ and ε₂ represent the relative permittivity of the P₁ and P₂ regions, respectively, while ε₀ denotes the vacuum permittivity.
Thermoelectric Transistor Cooling Performance
Thermoelectric Transistor Conversion Efficiency
The conversion efficiency (η) of a thermoelectric cooler is defined as:
Where η is the conversion efficiency, Qc denotes the heat absorption at the cold end, and P stands for the input electrical power.
When considering the coupling between the Seebeck effect and interface effects in NPN heterostructures, can be expressed as:
In the equation above, λP and λN represent the thermal conductivities of the P-type and N-type materials, respectively, while AP, lP and AN, lN are the cross-sectional areas and lengths of the P-type and N-type materials, respectively.
Cooling Temperature Difference
The input power P of the refrigeration device can be expressed as:
By substituting equations (39) and (41) into the equation of efficiency (40), we can obtain
Another important performance indicator of a thermoelectric cooler is the temperature difference that can be established across its two ends, that is
Using the thermal equilibrium equation of the cold end of the device (Equation (42)), the result can be obtained
When the cooler works with no external heat load (), the maximum temperature difference that the cooler can achieve can be obtained.
Calculation of
Since the thermoelectric transistor dissipates heat only at the forward bias and absorbs heat at the reverse bias, the heat absorbed at the forward bias portion, , equals the heat dissipated at the reverse bias portion, . This can be expressed as
Based on equations (48) and (51), we can determine , which is
During the calculation process, can be approximated as .
Calculation Methods and Process
The results of this work are computed using MATLAB R2020a. During the computation, is set equal to , and was fixed at 300K. Figure 4 illustrates the schematic of the optimization design and computational process. Using the trial-and-error method, first determine the values of , , and . Then, substitute the values of , , , and into formula (43). Under the condition of and , adjust the value of and calculate the corresponding values of , , and . Finally, obtain the allowable range of changes in , , , and . Based on the results above, the parameters , , and are optimized. The maximum temperature difference under different doping concentrations is calculated, respectively. After comparison, the optimal doping concentration was obtained.

Figure 4. Calculation flowchart of optimal doping concentration
Results and Discussion
In this work, the thickness of the thermoelectric transistor along the Y and Z directions is 0.1 mm, and the width of the N and N regions along the X direction is 0.06 mm. The high-temperature end temperature is maintained at 300 K, and it is assumed that the carrier concentrations of the emitter N1and the collector N2 are and , respectively, with . The performance parameters of other materials are shown in Table 1. In this calculation, since the maximum temperature difference is obtained when – at relatively low , the numerical simulation results for – are provided.
The calculation result of
Figure 5 (a), (b), (c), and (d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under N type region doping concentrations . From Figure 5 (a), under the same , as the voltage applied to the emitter increases, the collector voltage increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of also increase, as do the initial and maximum reverse bias voltages that need to be applied to the collector. From Figure 5 (b), it can be observed that the base thickness decreases as the voltage applied to the emitter and the base carrier concentration increases. From Figure 5 (c), when is constant, as increases, the temperature difference across the device decreases. When is constant, as increases, increases, and the maximum temperature difference across the device also increases with . In Figure 5 (d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Combining Figure 5 (a), (Figure 5b), (Figure 5c), and (Figure 5d), it can be seen that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . Comparing the maximum temperature differences under different conditions, the maximum temperature difference across the device increases from K at V, V, and to K at V, V, and . At these points, the base thickness is 0.36 nm and 0.16 nm, respectively.

Figure 5. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 6 (a), (Figure 6b), (Figure 6c), and (Figure 6d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under N type region doping concentrations . From Figure 6 (a), under the same , as the emitter voltage increases, the collector voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of increase, as do the initial and maximum reverse bias voltages that need to be applied to the collector. From Figure 6 (b), it can be observed that the base thickness decreases as the emitter voltage and base carrier concentration increases. From Figure 6 (c), when is constant, the temperature difference across the device decreases as increases. When is constant, increases with , and the maximum temperature difference across the device also increases with . In Figure 6 (d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Combining Figure 6 (a), (Figure 6b), (Figure 6c), and (Figure 6d), it can be seen that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . Comparing the maximum temperature differences under different conditions, the maximum temperature difference across the device increases from K at V, V, and to K at V, V, and . At these points, the base thickness is 0.98 nm and 0.17 nm, respectively.

Figure 6. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 7 (a), (Figure 7b), (Figure 7c), and (Figure 7d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under N type region doping concentration . From Figure 7 (a), under the same , as the emitter voltage increases, the collector voltage increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of also increase, as the initial and maximum reverse bias voltages required to be applied to the collector. From Figure 7 (b), it can be observed that the base thickness decreases as the emitter voltage and base carrier concentration increase. From Figure 7 (c), when is constant, the temperature difference across the device decreases as increases. When is constant, increases with , and the maximum temperature difference across the device also increases with . In Figure 7(d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Combining Figure 7 (a), (Figure 7b), (Figure 7c), and (Figure 7d), it can be concluded that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . Comparing under different conditions, the maximum temperature difference across the device increases from at , , and to at , , and . At these conditions, the base thickness is 1.96 nm and 0.19 nm, respectively.

Figure 7. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 8 (a), (Figure 8b), (Figure 8c), and (Figure 8d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under the condition of N type region doping concentration . From Figure 8 (a), under the same , as the emitter voltage increases, the collector voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of increase, as do the initial and maximum reverse bias voltages that need to be applied to the collector. From Figure 8 (b), it can be observed that the base thickness decreases as the emitter voltage and base carrier concentration increases. From Figure 8 (c), when is constant, the temperature difference across the device decreases as increases. Conversely, when is constant, increases with , and the maximum temperature difference also increases as increases. In Figure 8(d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Combining the information from Figure 8 (a), (Figure 8b), (Figure 8c), and (Figure 8d), it can be concluded that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . Comparing the maximum temperature differences under different conditions, the maximum temperature difference increases from K at V, V, and m to K at V, V, and m. At these conditions, the base thickness is 3.70 nm and 0.20 nm, respectively.

Figure 8. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 9 (a), (Figure 9b), (Figure 9c), and (Figure 9d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under the condition of N type region doping concentration m. From Figure 9 (a), under the same , as the emitter voltage increases, the collector voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of increase, as do the initial and maximum reverse bias voltages that need to be applied to the collector. From Figure 9 (b), it can be observed that the base thickness decreases as the emitter voltage and base carrier concentration increases. From Figure 9 (c), when is constant, the temperature difference across the device decreases as increases. Conversely, when is constant, increases with , and the maximum temperature difference also increases as increases. In Figure 9 (d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Combining the information from Figure 9 (a), (Figure 9b), (Figure 9c), and (Figure 9d), it can be concluded that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . Comparing the maximum temperature differences under different conditions, the maximum temperature difference increases from K at V, V, and m to K at V, V, and m. At these conditions, the base thickness is 5.58 nm and 0.22 nm, respectively.

Figure 9. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 10 (a), (Figure 10b), (Figure 10c), (Figure 10d) show the variation laws of collector voltage , base depletion layer thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under the condition of N type region doping concentration . From Figure 10 (a), under the same , as the emitter voltage increases, the collector voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of , as well as the initial and maximum reverse bias voltages required to be applied to the collector, also increase. From Figure 10 (b), it can be observed that the base thickness decreases as the emitter voltage and the base carrier concentration increases. From Figure 10 (c), when is constant, the temperature difference across the device decreases with increasing . When is constant, increases with increasing . Furthermore, the maximum temperature difference across the device first increases and then decreases as increases, reaching its maximum value when . From the comprehensive analysis of Figure 10 (a), (Figure 10b), (Figure 10c), and (Figure 10d), it can be concluded that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . By comparing the maximum temperature differences under different conditions, the maximum temperature difference across the device increases from at , , and , to at , , and , and then decreases to at , , and . The corresponding base thicknesses are 17.23 nm, 8.22 nm, and 5.53 nm, respectively.

Figure 10. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 11 (a), (Figure 11b), (Figure 11c), and (Figure 11d) show the variation laws of collector voltage , base region depletion layer thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration , under the condition of N type region doping concentration . From Figure 11(a), it can be observed that under the same , as the voltage applied to the emitter increases, the collector biot voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of , as well as the initial and maximum reverse bias voltages required to be applied to the collector, also increase. Furthermore, compared with Figure 5 to Figure 10, it can be found that as increases, the initial allowable value of increases, and the numerical range decreases rapidly. From Figure 11 (b), the base region thickness decreases as the voltage applied to the emitter and the base carrier concentration increases. From Figure 11 (c), when is constant, it can be observed that as increases, the temperature difference across the device decreases. When is constant, as increases, increases. Moreover, the maximum temperature difference across the device first increases and then decreases as increases, reaching its maximum value when , with . From Figure 11 (d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Integrating the information from Figure 11 (a), (Figure 11b), (Figure 11c), and (Figure 11d), it can be concluded that when is constant, the temperature difference across the device is maximized when is at its minimum allowable value, with corresponding minimum values of and , and . Comparing the maximum temperature differences under different conditions, the maximum temperature difference across the device increases from at , , and , to at , , and , and then decreases to K at V, V, and m. The corresponding base region thicknesses are 26.33 nm, 23.20 nm, and 6.74 nm, respectively.

Figure 11. (a), (b), (c), (d) as functions of and when m
The calculation result of m
Figure 12 (a), (b), (c), and (d) show the variation laws of collector voltage , base region thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under the condition of N type region doping concentration m. From Figure 12 (a), under the same , as the emitter voltage increases, the collector voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of increase, as do the initial and maximum reverse bias voltages that need to be applied to the collector. Comparing with Figure 5 to Figure 11, it can be observed that as increases, the initial allowable value of further increases, and its numerical range further decreases. From Figure 12 (b), the base region thickness decreases as the emitter voltage and base carrier concentration increases. From Figure 12 (c), when is constant, it can be observed that as increases, the temperature difference across the device decreases. Conversely, when is constant, as increases, increases. From the comprehensive analysis of Figure 12 (a), (Figure 12b), (Figure 12c) and (Figure 12d), it can be observed that when is constant, attains its minimum allowable value, resulting in the maximum temperature difference across the device terminals. Correspondingly, and also reach their minimum values, with . By comparing under different conditions, the maximum temperature difference across the device terminals increases from at , , and , to at , , and , and then decreases to at , , and . The corresponding base region thicknesses are , , and , respectively.

Figure 12. (a), (b), (c), (d) as functions of and when
The calculation result of
Figure 13 (a), (b), (c), and (d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under N type region doping concentrations . From Figure 13 (a), under the same , as the voltage applied to the emitter increases, the collector voltage increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of also increase, as do the initial and maximum reverse bias voltages that need to be applied to the collector. Comparing with Figures 5 to 12, it can be observed that as increases, the initial allowable value of further increases, and its numerical range further decreases. From Figure 13 (b), the base thickness decreases as the voltage applied to the emitter and the base carrier concentration increase. From Figure 13 (c), it can be observed that when is constant, the temperature difference across the device decreases as increases. When is constant, increases as increases, and the maximum temperature difference across the device decreases as increases. From Figure 13 (d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. Combining Figures 13 (a), (b), (c), and (d), it can be concluded that when is constant, the minimum allowable value of is taken, resulting in the maximum temperature difference across the device, with the corresponding and also taking their minimum values, and . Comparing the maximum temperature differences under different conditions, the maximum temperature difference across the device decreases from K at V, V, and m to K at V, V, and m. The corresponding base thicknesses are 42.26 nm and 8.73 nm, respectively.
The calculation result of m
Figure 14 (a), (b), (c), and (d) show the variation laws of collector voltage , base thickness , temperature difference across the device, and thermoelectric conversion efficiency with respect to emitter voltage and base carrier concentration under N type region doping concentrations m. From Figure 14 (a), under the same , as the emitter voltage increases, the collector voltage also increases. Additionally, as the base carrier concentration increases, the minimum and maximum allowable values of , as well as the starting and maximum reverse bias voltages required to be applied to the collector, also increase. Comparing with Figure 5 to Figure 13, it can be observed that as increases, the starting value of allowable further increases, and its numerical range further decreases. From Figure 14 (b), the base thickness decreases as the emitter voltage and base carrier concentration increases. From Figure 14 (c), it can be observed that when is constant, the temperature difference across the device decreases as increases; when is constant, increases as increases, and the maximum temperature difference across the device decreases as increases. In Figure 14 (d), the thermoelectric conversion efficiency decreases as increases when is constant and increases as increases when is constant. From the comprehensive analysis of Figure 14 (a), (Figure 14b), (Figure 14c), and (Figure 14d), it can be seen that when is constant, the temperature difference across the device is maximized when is set to its minimum allowable value, and the corresponding and are also minimized, with . By comparing the maximum temperature differences under different conditions, the maximum temperature difference across the device decreases from K at V, V, and m to K at V, V, and m. The corresponding base thicknesses are 50.05 nm and 9.53 nm, respectively.

Figure 13. (a), (b), (c), (d) as functions of and when m

Figure 14. (a), (b), (c), (d) as functions of and when m
Comprehensive Data Analysis
Figure 15 shows , , , and versus base region concentration when the N type region doping concentration is -. As shown in Figure 15 (e), when is -, the maximum value of has already appeared in the range of -. Therefore, in Figure 15 (e) to (Figure 15h), only the calculation results for are provided. In contrast, in Figure 15 (a) to (Figure 15d), the calculation results are presented when the value of ranging from to .

Figure 15. (a), (b), (c), and (d) versus base region concentration when –, and (e), (f), (g), and (h) versus base region concentration when –
From Figure 15 (a), when the N type region doping concentration is in the range of -, the N type region doping concentration increases, the minimum permissible base region carrier concentration decreases. For any given value, increases with the increase of , Moreover, as increases, increases rapidly at first, reaches a certain inflection point, and then increases slowly or even tends to stabilize. The maximum value of corresponding to all 5 values is obtained at . Besides changing with , also changes with , that is, changes simultaneously with both and . When , the maximum temperature difference reaches its maximum value of - K when , and the corresponding is - nm. When , reaches its maximum value of - K when , and the corresponding is - nm. When , reaches its maximum value of - K when , and the corresponding is - nm When , reaches its maximum value of - K when , and the corresponding is - nm. That is, as the N type region doping concentration increases, first increases, reaches a maximum, and then decreases. For example, starting from K at and , increases, reaching its maximum of K at and , and then begins to decrease, such as reaching a maximum of K at and . The maximum values of for N type region doping concentrations of , , , , and are K, K, K, K, and K, respectively. That is, the maximum temperature difference K is achieved at and , which is also the maximum temperature difference obtained in this study. Corresponding to K, V, V, and nm.
From Figure 15 (e), when the N type region doping concentration is in the range of $1.5$-, the decreases with the increasing of , and the corresponding value also decreases. Therefore, there is no calculation for under the condition of . Additionally, for different values, the law of changing with is different. When the N type region doping concentration is in the range of -, varies with in a parabolic pattern. As increases, first increases, reaches its maximum at the vertex of the parabola, and then gradually decreases. Moreover, the maximum value of decreases as the doping concentration in the N region increases. When is , and the maximum value of is K and the corresponding parameters are , V, V, and nm. When the N type region doping concentration increases to , the extremum point of shifts to , achieving a value of K, with corresponding parameters V, V, and nm. When , the extremum point of occurs at , reaching K, with V, V, and nm. For N type region doping concentrations $N_d \ge , decreases as increases, that is, attains its maximum value at the minimum value of . When , the maximum value of is 122.95 K, with , V, V, and nm. When , the maximum value of is 114.32 K, and the corresponding parameters are , V, V, and nm.
From Figure 15 (b), (Figure 15c), (Figure 15f), and (Figure 15g), the externally applied voltages and at the emitter and collector increase with the base thickness and base concentration , and also increase with the emitter and collector concentrations . The reason is that according to equations (21) and (24), the interface voltage of the PN junction increases as carrier concentration increases, thus requiring a larger external voltage. From Figure 15 (d) and (Figure 15h), it can be observed that the base thickness decreases with increasing base concentration and increases with increasing emitter and collector concentrations , which aligns with equations (38) and (39).
Conclusion
In this study, an NPN junction thermoelectric transistor cooling device model composed of N-type and P-type is constructed. By separately regulating the externally applied voltages on the emitter and collector, and coupling the transistor effect, thermoelectric effect, and interface effect within the NPN heterostructure, the temperature at the forward-biased end (emitter) increases while releasing heat, and the temperature at the reverse-biased end (collector) decreases while absorbing heat, thereby forming a thermoelectric transistor refrigerator and enhancing its cooling performance. Using the equivalent circuit method to convert the PNP heterostructure into a common-base circuit of a junction thermoelectric transistor, When the doping concentration in the N region is – and the doping concentration in the P region is –, the effects of different applied voltages on the maximum temperature difference between the two ends of the thermoelectric transistor device and the base region thickness are investigated. The results are as follows:
Under different N type region doping concentrations and base (P type region) doping concentrations , as the voltage applied to the emitter increases, the collector voltage , the base depletion layer thickness , and the thermoelectric conversion efficiency all increase, while the temperature difference between the two ends of the thermoelectric transistor decreases;
When the N type region doping concentration is in the range of –, increases rapidly at first with increasing , reaches a certain inflection point, then increases slowly or even tends to stabilize. As the N type region doping concentration increases, first increases, reaches a maximum at , and then decreases. When is greater than , decreases as increases. When increases from to , decreases from 176.51 to 114.32 K. When is in the range of –, varies with in a parabolic pattern, where increases with increasing , reaches its maximum at the vertex of the parabola, and then decreases slowly. When , decreases as increases, the maximum value of is obtained at the minimum value of . Overall, as the N type region doping concentration increases, the minimum allowable base carrier concentration decreases.
The maximum temperature difference obtained is K under the conditions of emitter voltage V, collector voltage V, P type region
carrier concentration , and N type region carrier concentration . However, the corresponding base width is not only difficult to achieve but also poses a risk of breakdown due to its extremely small size. Therefore, in actual production, suitable doping concentrations and external voltages need to be selected based on the actual process and the results of the optimized design. Even if the base region width is limited to the ?nm level, when the carrier concentration in the N region () is , the carrier concentration in the base region () is , the emitter voltage is , the collector voltage is , and the base region width () is , the maximum temperature difference that can be obtained is approximately . Therefore, the thermoelectric transistor device, which combines the thermoelectric effect with transistor technology can effectively enhance the maximum temperature difference that thermoelectric cooling technology can achieve.
Acknowledgments
This work was financially supported by The National Key Research and Development Program of China [Grand No. 2017YFF0204706] and the Fundamental Research Funds for the Central Universities [Grant No. FRF-MP-18-005 and FRF-MP-19-005].
References
- [1]PELТIER J. Nouvelles experiences sur la caloricite des courants electrique[J]. Annales de Chimie et de Physique, 1834,56: 371-386.
- [2]J. Mao, G. Chen, Z. Ren. Thermoelectric cooling materials. Nat. Mater. 2021, 20, 454.
- [3]J. Yang, G. Li, H. Zhu, N. Chen, T. Lu, J. Gao, L. Guo, J. Xiang, P. Sun, Y. Yao, R. Yang, H. Zhao. Next-generation thermoelectric cooling modules based on high-performance Mg₃(Bi, Sb)₂ material. Joule 2022, 6, 193.DOI
- [4]Y. Xiao, H. Wu, H. Shi, L. Xu, Y. Zhu, Y. Qin, G. Peng, Y. Zhang, Z. H. Ge, X. Ding, L. D. Zhao. High-Ranged ZT Value Promotes Thermoelectric Cooling and Power Generation in n-Type PbTe. Adv. En-ergy Mater. 2022.DOI
- [5]Sun J-C, Su X-L, Yan Y-G, Liu W, Tan G-J, Tang X-F. Enhancing thermoelectric performance of n-type PbSe through forming solid solution with PbTe and PbS. ACS Appl. Energy Mater. 2020; 3: 2-8 (2020).DOI
- [6]Perumal S, Samanta M, Ghosh T, Shenoy U-S, Bohra AK, Bhattacharya S, Singh A, Waghmare U V, Biswas K. Realization of high thermoelectric figure of merit in GeTe by complementary Co-doping of Bi and In. Joule. 2019; 3: 2565-80.DOI
- [7]G.D. Mahan Introduction to thermoelectrics. APL Mater. 4 (2016), 104806.
- [8]J.T. Wei, L.L. Yang, Z. Ma, P.S. Song, M.L. Zhang, J. Ma, F.H. Yang, X.D. Wang, J. Mater. Review of current high-ZT thermoelectric materials. Sci. 55 (2020) 12642–12704.
- [9]M. Bala, A. Masarrat, V. Kumar, S. Ojha, K. Asokan, S. Annapoorni. Effect of thermal annealing on thermoelectric properties of BiₓSb₂−ₓTe₃ thin films grown by sputtering. J. Appl. Phys. 127 (2020), 245108.DOI
- [10]I. Malik, T. Sribastava, K.K. Surthi, C. Gayner, K.K. Kar, Enhanced thermoelectric performance of n-type Bi₂Te₃ alloyed with low cost and highly abundant sulfur, Mater. Chem. Phys. 255 (2020), 123598.DOI
- [11]J.L. Gao, T. Mao, T. Lv, Z.M. Li, G.Y. Xu. Thermoelectric performance of n-type (PbTe)₁₋ₓ(CoTe)ₓ composite prepared by high pressure sintering method. J. Mater. Sci.-Mater. El 29 (2018) 5327–5336.DOI
- [12]N.T. Hung, E.H. Hasdeo, A.R.T. Nugraha, M.S. Dresselhaus, R. Saito. Quantum Effects in the Thermoelectric Power Factor of Low-Dimensional Semiconductors. Phys. Rev. Lett. 117 (2016), 036602.
- [13]L.P. Hu, T.J. Zhu, X.H. Liu, X.B. Zhao. Point Defect Engineering of High-Performance Bismuth-Telluride-Based Thermoelectric Materials. Adv. Funct. Mater. 24 (2014) 5211–5218.DOI
- [14]B.C. Qin, Y. Xiao, Y.M. Zhou, L.D. Zhao. Thermoelectric transport properties of Pb–Sn–Te–Se system. Rare Metals 37 (2018) 343–350.
- [15]S. Bathula, M. Jayasimhadri, N. Singh, A.K. Srivastava, J. Pulikkotil, A. Dhar, R. C. Budhani. Enhanced thermoelectric figure-of-merit in spark plasma sintered nanostructured n-type SiGe alloys. Appl. Phys. Lett. 101 (2012), 213902.DOI
- [16]C. Lan, A.J. Minnich, G. Chen, Z.F. Ren. Enhancement of Thermoelectric Figure-of-Merit by a Bulk Nanostructuring Approach. Adv. Funct. Mater. 20 (2010) 357–376.
- [17]J.R. Sootsman, R.J. Pcionek, H.J. Kong, C. Uher, M.G. Kanatzidis. Strong Reduction of Thermal Conductivity in Nanostructured PbTe Prepared by Matrix Encapsulation. Chem. Mater. 18 (2006) 4993.DOI
- [18]A. Gueguen, P.F.P. Poudeu, C.P. Li, S. Moses, C. Uber, J.Q. He, V. Dravid, K. A. Paraskevopoulos, M.G. Kanatzidis. Thermoelectric Properties and Nanostructuring in the p-Type Materials NaPb₁₈₋ₓSnₓMTe₂₀ (M=Sb, Bi). Chem. Mater. 21 (2009) 1683–1694.DOI
- [19]G.H. Zhu, H. Lee, Y.C. Lan, X.W. Wang, G. Joshi, D.Z. Wang, J. Yang, D. Vashaee, H. Guilbert, A. Pillitteri, M.S. Dresselhaus, G. Chen, Z.F. Ren. Increased Phonon Scattering by Nanograins and Point Defects in Nanostructured Silicon with a Low Concentration of Germanium. Phys. Rev. Lett. 102 (2009), 196803.
- [20]L.D. Hicks, M.S. Dresselhaus. Effect of quantum-well structures on the thermoelectric figure of merit. Phys. Rev. B 47 (1993) 12727.
- [21]R. Venkatasubramanian, E. Siivola, T. Colpitts, B. O’Quinn. Thin-film thermoelectric devices with high room-temperature figures of merit. Nature 413 (2001) 597–602.
- [22]T.C. Harman, M.P. Walsh, B.E. Laforge, G.W. Turner, J. Electron. Nanostructured thermoelectric materials. Mater. 34 (2005) L19–L22.
- [23]A.I. Boukai, Y. Bunimovich, J. Tahir-Kheli, J.K. Yu, W.A. Goddard, J.R. Heath. Silicon nanowires as efficient thermoelectric materials. Nature 451 (2008) 168–171.DOI
- [24]B. Hinterleitner, I. Knapp, M. Poneder, Y.P. Shi, H. Muller, G. Eguchi, C. Eisenmenger-Sittner, M. Stoger-Pollach, Y. Kakefuda, N. Kawamoto. Thermoelectric performance of a metastable thin-film Heusler alloy. Nature 576 (2019) 85.DOI
- [25]曹海山. 热电制冷技术进展与展望[J]. 制冷学报, 2022, 43 (04): 26–34.
- [26]Feliczo J-C, Uenuma M, Fujii M-N, Uraoka Y. Improved Thermoelectric Power Factor of InGaZnO/SiO₂ Thin Film Transistor via Gate-Tunable Energy Filtering. IEEE. Electron. Device. L. 2021; 42: 1236-1239.DOI
- [27]Bejenari I, Kantser V, Balandin AA. Thermoelectric properties of electrically gated bismuth telluride nanowires. Phys. Rev. B. 2010; 81: 075316.DOI
- [28]Qin D-L, Pan F, Zhou J, Xu Z-B, Deng Y. High ZT and performance controllable thermoelectric devices based on electrically gated bismuth telluride thin films. Nano Energy. 2021; 89: 106472.DOI
- [29]Y ang F, Wu J, Suwardi A, Zhao Y-S, Laing B-Y, Jiang J, Xu J-W, Chi D-Z, Hippalgaonkar K, Lu J-P , Ni, Z-H. Gate-Tunable Polar Optical Phonon to Piezoelectric Scattering in Few-Layer Bi₂O₂Se for High-Performance Thermoelectrics. Adv. Mater. 2021; 4: 33.DOI
- [30]Wu X-M, Gao G-Y, Hu L, Qin D. 2D Nb2SiTe4 and Nb2GeTe4: promising thermoelectric figure of merit and gate-tunable thermoelectric performance. Nanotechnology. 2021; 32: 245203.DOI
- [31]Bohang Nan, Tao Guo, Hao Deng, Guangbing Zhang, Ran Shi, Jiakai Xin, Chen Tang, Guiying Xu, Output performance improvement for thermoelectric transistor with the consideration of the Thomson effect and geometry optimization, Applied Energy, 2024, 357, 122523DOI
- [32]Nan B H, Xu G Y, Yang Q X, Zhang B, Zhou X J, Innovative design and optimized performance of thermoelectric transistor driven by the Seebeck effect, Energy Conversion and Management[J]. 2023, 283, 116880DOI
- [33]Nan Bohang, Xu Guiying, Liu Wu-Ming, Yang Quanxin, Zhang Bin, Dong Yuan, Tie Jian, Guo Tao, Zhou Xiaojing, High thermoelectric performance of PNP abrupt heterostructures by independent regulation of the electrical conductivity and Seebeck coefficient, Materials Today Communications, 2022, 31: 103343.DOI
- [34]M. Zebarjadi, A. Shakouri, K. Esfarjani. Thermoelectric transport perpendicular to thin-film heterostructures calculated using the Monte Carlo technique Phys. Rev. B 74(2006), 195331.DOI
- [35]V.N. Agarev. Nonstationary thermoelectric power in multilayered structures with p-n junctions. Semiconductors 31 (1997) 784.DOI
- [36]K. Kim, J. Park, S.U. Kim, O. Kwon, J.S. Lee, S.H. Park, Y.K. Choi. Thermopower profiling of a silicon p-n Junction. Appl. Phys. Lett. 90 (2007) 43107.
- [37]X.Y. Wang, H.J. Wang, B. Xiang, L.W. Fu, H. Zhu, D. Chai, B. Zhu, Y. Yu, N. Gao, Z. Y. Huang, F.O. Zu. Thermoelectric Performance of Sb₂Te₃-Based Alloys is Improved by Introducing PN Junctions. ACS Appl. Mater. Interfaces 10 (2018) 23277.DOI
- [38]I.I. Balmush, Z.M. Dashevskii, A.I. Kasiyan. Potential Distribution in Semiconductor Structure with p−n Junction in the Presence of Temperature Gradient. Fiz. i Tekh. Poluprovodn. 20 (1986)1541 (In Russian).
- [39]Z.M. Dashevsky, S. Ashmontas, L. Vingelis, I. Gradauskas, Al. Kasian. The thermoelectric power on p-n junction. In 15th Int. Conf. on Thermoelectrics. IEEE. 336 (1996).
- [40]A.A. Zakhidov, Y.I. Ravich, D.A. Pchenoy-Severin. Thermopower enhancement and optimal ZT in p-n junction arrays. In 18th Int. Conf. on Thermoelectrics. IEEE. 193 (1999).
- [41]Y.I. Ravich, D.A. Pshenai-Severin. Thermoelectric figure of merit of a p-n junction. Semiconductors 35 (2001) 1161–1165.
- [42]Chen Z W, Jian Z Z, Li W, et al. Lattice dislocations enhancing thermoelectric PbTe in addition to band convergence [J]. Advanced Materials, 2017, 29(23): 1606768.DOI
- [43]He J, Xu J T, Liu G Q, et al. Enhanced thermopower in rock-salt SnTe–CdTe from band convergence [J]. RSC Advances, 2016, 6(38): 32189-32192.DOI
- [44]V. A. Kulbachinskii, M. Inoue, M. Sasaki, H. Negishi, W. X. Gao, K. Takase, Y. Giman, P. Lostak, and J. Horak. Valence-band changes in Sb₂−xInxTe₃ and Sb₂Te₃−ySey by transport and Shubnikov–de Haas effect measurements. Phys. Rev. B 50, 16921pubmed.ncbi.nlm.nih.gov/9980190
- [45]Xu G-Y, Ren P, Lin T, Wu X-F, Zhang Y-H, Niu S-T, Bailey TP. Mechanism and application method to analyze the carrier scattering factor by electrical conductivity ratio based on thermoelectric property measurement. J. Appl. Phys. 2018; 123: 015101.
- [46]Wang S-Y, Tan G-J, Xie W-J, Zheng G, Li H, Yang J-H, Tang X-F. Enhanced thermoelectric properties of Bi₂(Te₁−xSex)₃-based compounds as n-type legs for low temperature power generation. J. Mater. Chem. 2012; 22: 20943.DOI
- [47]Dai W, Liu W-K, Yang J, Xu C, AlabastrI A, Liu C, Nordlander P, Guan Z-Q, Xu H-X. Giant photothermoelectric effect in silicon nanoribbon photodetectors. Light Sci. Appl. 2020; 9: 120.DOI
- [48]Ge Y, He K, Xiao L-H, Yuan W-Z, Huang S-M. Geometric optimization for the thermoelectric generator with variable cross-section legs by coupling finite element method and optimization algorithm. Renew Energ. 2022; 183: 294.DOI
- [49]Fu D, Levander A-X, Zhang R, Ager III J-W, Wu J. Electrothermally driven current vortices in inhomogeneous bipolar semiconductors. Phys. Rev. B. 2011; 84: 045205.DOI
- [50]Lv T, Li Z-M, Yang Q-X, Benton A, Zheng H-T, Xu G-Y. Synergistic regulation of electrical-thermal effect leading to an optimized thermoelectric performance in Co doping n-type Bi₂(Te₀.₉₇Se₀.₀₃)₃. Intermetallics. 2020; 118: 106683.DOI
- [51]Chavez R, Angst S, Hall J, Stoetzel J, Kessler V, Bitzer L, Maculewicz F, Benson N, Wiggers H, Wolf D, Schierning G, Schmechel R. High Temperature Thermoelectric Device Concept Using Large Area PN Junctions. J. Electron. Mater. 2014; 43: 2376.DOI