A method of measuring the thermovoltaic effect in heterogeneous media with gradient doping impurity distributions producing gradient carrier distributions has been proposed. Iron doped zinc oxide specimens have been produced using ion beam sputtering on thin foil tantalum substrates for thermovoltaic effect measurements, glass-ceramic substrates for Hall measurements and silicon substrates for structural study. The doping impurity concentration х_Fe in the specimens has been varied from 0.34 to 4.18 at.%. X-ray phase analysis has shown that all the specimens have a hexagonal zinc oxide crystal structure. The films have preferential [002] orientation. The carrier concentration in the experimental specimen layers according Hall data obtained on an ECOPIA 5500 measurement system in a 0.5 T DC magnetic field has varied in the 10¹⁶–10²⁰ cm^-3 range. The specimens have an n-type conductivity. Thermovoltaic measurements have been carried out for two-layered iron doped zinc oxide specimens with different carrier and iron doping impurity concentrations using the method proposed. The maximum thermovoltaic response (U ~ 80 μV) has been observed in the two-layered thin-film specimen with the carrier concentration difference between the layers (Δn ≈ 2∙10³ cm^-3). The observed saturation of the thermovoltaic response has been attributed to the establishment of dynamic equilibrium between carrier diffusion from the high carrier concentration layer to the low carrier concentration layer and carrier drift due to internal electric field.

thermovoltaic effect, zinc oxide, doping impurity

Thermovoltaic converters have recently attracted great interest in electric power generation. The thermovoltaic effect (TVE) is spontaneous generation of electromotive force (emf) in gradient-doped semiconductor materials upon uniform heating [1–3]. The fundamental difference between the TVE and the classical Seebeck effect is that the TVE originates not from a temperature gradient but from a chemical potential gradient produced by an impurity concentration gradient and hence a carrier concentration gradient. Then thermal energy is converted to electric power upon uniform specimen heating, i.e., in the absence of a temperature gradient.

In Russia, this effect was first discovered and studied for semiconducting samarium sulfide [3–11]. The TVE mechanism according to the Authors [10] is as follows. Excess Sm²⁺ ions in interstitial positions of the lattice generate a shallow E_i ~ 45 meV donor level with a high carrier concentration, n ~ 10²⁰÷10²¹ cm^-3. Temperature growth to a critical level near 400 K thermally activates the carriers at that level thus increasing their concentration in the conduction band. At T ~ 400 K, the concentration n becomes sufficient for the complete delocalization of electrons from the E_i level in some local regions due to the screening of the electric potential of the impurity. This delocalization is accompanied by an abrupt surge of the carrier concentration in the conduction band causing Mott’s transition in the abovementioned local regions. Those nonequilibrium electrons have a concentration gradient which produces a chemical potential gradient which in turn triggers their diffusion from high carrier concentration regions to low carrier concentration regions thus producing a voltage. This complex process is maintained by permanent heating and can continue for an arbitrary long time. Later on, the presence of thermovoltaic response was also confirmed for homogeneous junctions in zinc oxide doped with variable valence impurities such as iron and copper [1].

As indicated earlier [13], TVE based thermoelectric heat power converters should have numerous advantages upon uniform heating of a semiconducting material: no need to produce a temperature gradient, higher conversion efficiency, lower specific weight etc. From the physical viewpoint, a temperature gradient required for thermoelectric heat power conversion in TVE based generators can be replaced for an impurity concentration gradient (e.g. its inhomogeneous distribution) in a semiconducting thermoelectric material of n- or p-conductivity type. Gradient TVE structures can be synthesized in bulk semiconductors but batch fabrication of batteries might face problems. However, the synthesis of TVE based thermoelectric heat power converters can be implemented through a relatively small number of process operations if a thin-film technology is used.

Studies of TVE in gradient materials face difficulties in maintaining a constant temperature in the whole test specimen bulk for avoiding the effect of thermo-emf on the measurement results. Taking into account the above, a TVE measurement method was proposed and TVE was measured in two-layered zinc oxide structures with different carrier concentrations in the layers. Zinc oxide was chosen as the test material since polycrystalline zinc oxide has high thermo-emf, low thermoelectric quality (ZT) due to low carrier concentrations (and hence low conductivity) and high heat conductivity coefficient [14–16]. The synthesis and study of thin-film zinc oxide is furthermore of great scientific importance since due to its good transparency and high electron mobility it finds new applications in transparent electrodes for LCDs [14], gas sensors [15], thin-film transistors and LEDs [16] and other devices [17]. The synthesis and stabilization of nanostructures with preset properties is a complex task since nanosized materials have large surface area and high reactivity. Therefore the choice of a synthesis process plays an important role in the formation of nanostructures with required parameters such as crystallite sizes and their distribution, shape, homogeneity etc. Correct process choice can deliver practically applicable materials.

There are various zinc oxide thin film and coating synthesis methods: chemical vacuum deposition [21], electrochemical deposition [22], molecular beam epitaxy [23], pyrolysis [24], vapor phase deposition with thermal [25], pulse laser [26], magnetron [27] and carbothermal sputtering [28], pyrolysis of thiourea coordination compound aerosols onto heated substrates (spray method) [29], low pressure chemical vapor phase deposition (CVD) [30].

Below we consider ion beam sputtering synthesis of Zn_1-xFe_xO films with different carrier concentrations, TVE measurement method developed herein and experimental data on the structural, optical and thermal properties that are important for practical applications.

The test specimens were synthesized by ion beam sputtering in an UVN-2M vacuum plant described in detail earlier [31]. The substrates were thin foils of tantalum (two-layered specimens to TVE measurement), glass-ceramic (single-layered specimens to Hall measurements) and silicon to structural studies. The bottom layers of the two-layered structures with different iron contents were synthesized by sputtering a water-cooled target consisting of ZnO ceramic base with nonuniformly arranged Fe₂O₃ plates. This method allows synthesizing specimens with different doping element (Fe) concentrations during single deposition process cycle. The top layers of the two-layered structures were deposited from pure ZnO ceramic target. The thickness of the films as measured with a MII-4 interferometer was ~0.75 μm per each layer. The Fe content х_Fe in the specimens as measured using the selected area electron probe method ranged from 0.34 to 4.18 at.%. It should be noted that the pure ZnO layers contained small Fe quantities due to specific issues of the ion beam sputtering process.

The structure and phase composition of the specimens were studied using X-ray phase analysis. The measurements were carried out on a Bruker D2 Phaser diffractometer (λ_CuKα1 = 0.154 nm). The diffraction patterns were analyzed with the DIFFRAC.EVA 3.0 software and the ICDD PDF Release 2012 database [32].

The carrier concentration in the specimen layers was measured using the Hall method on an ECOPIA 5500 measurement system in a 0.5 T DC magnetic field. The Hall specimens were squared, 8×8 mm², with ultrasonically soldered indium contacts.

X-ray phase analysis data for ZnO:Fe films with different Fe concentrations are shown in Fig. 1. Analysis of the diffraction patterns showed that all the specimens had a hexagonal zinc oxide structure. Similar data were obtained earlier for iron doped zinc oxide [33]. The films had preferential [002] orientation.

Figure 1.

X-ray diffraction patterns of ZnO:Fe films with different Fe concentrations x_Fe (at.%): (1) 2.85, (2) 3.78, (3) 4.12, (4) 4.91

The thermovoltaic effect was measured on a Netzsch SBA 458 thermo-emf coefficient and specific electrical conductivity measurement system. Figure 2 shows schematic of the measurement cell. The cell is placed in an electric furnace for measurements. A two-layered zinc oxide specimen is installed in the center of the measurement cell on ceramic supports. One layer of the specimen is heavily ion doped (brown). Micro-heaters (red) are installed inside the Al₂O₃ ceramic specimen supports at two sides. Two alumel-chromel (K-type) thermocouples and two current probes for Seebeck coefficient and electrical conductivity measurement respectively are connected to the bottom specimen surface. The thermocouples and the current probes are secured in the ceramic specimen supports. Springs exert a constant load on the thermocouples and the current probes in the cold part of the measurement cell. A pressure pad and an additional load press the specimen to the specimen supports with a constant force.

The entire measurement procedure is automatic and is implemented as follows. The specimen is heated in accordance with the preset stepwise temperature program. Upon reaching the required isothermal exposure temperature T the Seebeck coefficient was measured and then the TVE was measured following the method described below. A temperature gradient was produced for TVE measurement by heating the right-hand side and cooling the left-hand side of the specimen. Then the U₁ = f (ΔT) curve was recorded where U₁ is the voltage measured relative to the respective branches of the chromel thermocouples. Then a temperature gradient was produced by heating the left-hand side and cooling the right-hand side of the specimen. Then the U₂ = f (ΔT) curve was recorded where U₂ is the voltage measured relative to the respective branches of the alumel thermocouples. The next stage was plotting linear functions for the resultant U = f (ΔT) curves (Fig. 3). The TVE voltage was calculated as the arithmetical mean of the data at the ΔT = 0 point of the U₁ and U₂ linear functions of ΔT, i.e., at a zero temperature difference (Fig. 3).

The TVE measurement process described above was used at all the preset temperature steps. It can be seen from Fig. 3 that if this method is used for measuring a uniform specimen (the Ta foil) without temperature gradients, the voltages at the respective thermocouple branches are zero (Fig. 3, curves 3 and 4), i.e., there is no TVE response. The situation is quite different for a two-layered ZnO/ZnO:Fe specimen (Fig. 4, curves 1 and 2): the respective thermocouple branches intersect at one point U = 2.9∙10^-5 V at a temperature difference of ΔT = 0, i.e., the specimen exhibits the TVE.

Figure 2.

Schematic of measurement cell

Figure 3.

Illustration of TVE response measurement in two-layered thin-film zinc oxide specimens: (1, 2) voltage relative to branches of chromel (U₁) and alumel (U₂) thermocouples, respectively (layer 1 (bottom) x_Fe = 0.47 at.% and layer 2 (top) x_Fe = 4.18 at. %); (3, 4) voltage relative to branches of chromel (U₁) and alumel (U₂) thermocouples, respectively, for a uniform specimen (Ta foil). Isothermal exposure temperature T = 373 K

Figure 4.

TVE response voltage as a function of temperature for two-layered thin-film ZnO/ZnO:Fe specimens with different iron concentrations: (1) specimen 1: layer 1 x_Fe = 0.34 at.%, layer 2 x_Fe = 4.00 at.%; (2) specimen 2: layer 1 x_Fe = 0.47 at.%, layer 2 x_Fe = 4.18 at.%; (3) specimen 3: layer 1 x_Fe = 0.55 at.%, layer 2 x_Fe = 2.18 at.%

Figure 4 shows TVE response voltage as a function of temperature for two-layered specimens the top layers of which contained different iron concentrations (see Fig. 2). With an increase in temperature the TVE response voltage grows due to an increase in the iron concentration in the undoped (top) layer. At about 423 K the first and second specimens undergo saturation whereas the third specimen (with lower iron concentration in the top layer) is not saturated. To understand the cause of this difference in TVE voltage behavior we measured Hall temperature curves for separate layers of the multilayered structures and calculated the difference between the carrier concentrations in the layers Δn for each specimen at 423 K (Fig. 5). The difference was Δn₁ ≈ 4.5∙10¹⁹ cm^-3, Δn₂ ≈ 7.1∙10¹⁹ cm^-3 and Δn₃ ≈ 1.0∙10²⁰ cm^-3 for Specimens 1, 2 and 3, respectively. Furthermore, TVE response voltage was studied as a function of temperature for two-layered thin-film Specimen 3 after sequential heating to 498 and 623 K (Fig. 6).

Without dwelling upon the main regularities of the observed TVE changes or causes of TVE growth with temperature and carrier concentration, one can state that the carrier concentration difference between the layers at 423 K is the greatest for Specimen 3 with the lowest iron concentration (2.18 at.%) in the top (heavier doped) zinc oxide layer (Fig. 5b). This seems to be the cause of the higher TVE voltage in that specimen. Analysis of the TVE response voltage vs temperature curves for the two-layered thin-film Specimen 3 (Fig. 6) showed that the TVE voltage decreased after sequential heating to 498 and 623 K, same as for Specimens 1 and 2, and sees saturation. These regularities can be accounted for as follows. Uniform heating of a gradient-doped semiconductor specimen triggers carrier diffusion from the higher carrier concentration layer to the lower carrier concentration layer.

Carrier diffusion produces an inner electric field in the specimen causing carrier drift opposite to the carrier diffusion direction. The resultant TVE response is determined by equilibrium between the carrier diffusion and drift currents, i.e., the TVE voltage sees saturation. A decrease in the TVE voltage as a result of sequential heating can be caused by the fact that the time of one measurement which is ~12 h is sufficient for structural relaxation in the specimens, e.g. the number of intrinsic ZnO defects can decrease and the metastable solid solution in the top specimen layer can decompose.

The activation energy of the thermovoltaic response in the synthesized structures was estimated by plotting the temperature functions in the ln U = f (1/T) coordinates (Fig. 7). The initial graph sections (see Fig. 7) are linear and the TVE activation energy was estimated from them to be U = 0.010 ± 0.005 eV. These TVE activation energy data can be logically attributed to carrier mobility fluctuations or hopping conductivity, as observed earlier [34]. Further study of this effect for gradient structures in zinc oxide or other semiconductors will solve a number of questions regarding the physics of this new and attractive phenomenon.

Thus, the origin of the observed TVE is the same (chemical potential gradient) as that of thermo-emf. The difference is that the temperature gradient that causes thermo-emf in specimens is replaced for a doping impurity concentration gradient which generates a thermovoltaic effect voltage even at a zero temperature gradient.

Figure 5.

Carrier concentration as a function of temperature at 300–550 K for bottom and top ZnO doped layer in test specimens (а) 1, (b) 2, (c) 3: (а) 1: layer 1 x_Fe = 0.34 at.%; 2: layer 2 x_Fe = 4.00 at.%; (b) 1: layer 1 x_Fe = 0.47 at.%; 2: layer 2 x_Fe = 4.18 at.%; (c) 1: layer 1 x_Fe = 0.55 at.%; 2: layer 2 x_Fe = 2.18 at.%. Points are experimental data; curves are second order polynomial approximation. Dashed line marks T = 423 K

Figure 6.

Thermovoltaic response as a function of temperature for two-layered thin-film ZnO/ZnO:Fe Specimen 3 after sequential heating of (1, 3) top and (2, 4) bottom layers to (1, 2) 498 and (3, 4) 623 K

Figure 7.

Logarithmic TVE voltage as a function of inverse temperature for thin ZnO/ZnO:Fe films: (1) Specimen 1: layer 1 x_Fe = 0.34 at.%, layer 2 x_Fe = 4.00 at.%); (2) Specimen 2: layer 1 x_Fe = 0.47 at.%, layer 2 x_Fe = 4.18 at.%; (3) Specimen 3: layer 1 x_Fe = 0.55 at.%, layer 2 x_Fe = 2.18 at.%

A TVE measurement technique for heterogeneous media with doping impurity concentration gradients producing carrier concentration gradients was developed. The TVE in two-layered thin-film zinc oxide specimens with different iron doping impurity concentrations was studied. The strongest TVE response was observed in the specimen with the greatest carrier concentration difference between the layers. The observed TVE saturation was attributed to the establishment of equilibrium between carrier diffusion from the high carrier concentration layer to the low carrier concentration layer and carrier drift due to inner electric field.

This work was carried out with support from the Russian Research Foundation, Project No. 24-29-20099.